Needed for backwards compatibility with old Java API
New features for first camera 2 release (API1)
Needed for useful RAW image processing and DNG file support
Entry is only used by camera device legacy HAL 2.x
Entry is required for full hardware level devices, and optional for other hardware levels
Entry is required for the depth capability.
Entry is required for the YUV or PRIVATE reprocessing capability.
Entry is required for logical multi-camera capability.
Entry is required for devices with HEIC (High Efficiency Image Format) support.
Entry is under-specified and is not required for now. This is for book-keeping purpose,
do not implement or use it, it may be revised for future.
android.util.Pair<Float,Float>
android.util.Pair<Double,Double>
android.graphics.Rect
android.util.Size
String
boolean
int
android.hardware.camera2.params.StreamConfigurationMap
android.hardware.camera2.params.StreamConfiguration
android.hardware.camera2.params.RecommendedStreamConfiguration
android.hardware.camera2.params.StreamConfigurationDuration
android.hardware.camera2.params.Face
android.hardware.camera2.params.MeteringRectangle
android.util.Range<Float>
android.util.Range<Integer>
android.util.Range<Long>
android.hardware.camera2.params.ColorSpaceTransform
android.hardware.camera2.params.RggbChannelVector
android.hardware.camera2.params.BlackLevelPattern
int
android.util.SizeF
android.graphics.Point
android.hardware.camera2.params.TonemapCurve
android.hardware.camera2.params.LensShadingMap
android.location.Location
android.hardware.camera2.params.HighSpeedVideoConfiguration
android.hardware.camera2.params.ReprocessFormatsMap
android.hardware.camera2.params.OisSample
android.hardware.camera2.params.MandatoryStreamCombination
TRANSFORM_MATRIX
Use the android.colorCorrection.transform matrix
and android.colorCorrection.gains to do color conversion.
All advanced white balance adjustments (not specified
by our white balance pipeline) must be disabled.
If AWB is enabled with `android.control.awbMode != OFF`, then
TRANSFORM_MATRIX is ignored. The camera device will override
this value to either FAST or HIGH_QUALITY.
FAST
Color correction processing must not slow down
capture rate relative to sensor raw output.
Advanced white balance adjustments above and beyond
the specified white balance pipeline may be applied.
If AWB is enabled with `android.control.awbMode != OFF`, then
the camera device uses the last frame's AWB values
(or defaults if AWB has never been run).
HIGH_QUALITY
Color correction processing operates at improved
quality but the capture rate might be reduced (relative to sensor
raw output rate)
Advanced white balance adjustments above and beyond
the specified white balance pipeline may be applied.
If AWB is enabled with `android.control.awbMode != OFF`, then
the camera device uses the last frame's AWB values
(or defaults if AWB has never been run).
The mode control selects how the image data is converted from the
sensor's native color into linear sRGB color.
When auto-white balance (AWB) is enabled with android.control.awbMode, this
control is overridden by the AWB routine. When AWB is disabled, the
application controls how the color mapping is performed.
We define the expected processing pipeline below. For consistency
across devices, this is always the case with TRANSFORM_MATRIX.
When either FULL or HIGH_QUALITY is used, the camera device may
do additional processing but android.colorCorrection.gains and
android.colorCorrection.transform will still be provided by the
camera device (in the results) and be roughly correct.
Switching to TRANSFORM_MATRIX and using the data provided from
FAST or HIGH_QUALITY will yield a picture with the same white point
as what was produced by the camera device in the earlier frame.
The expected processing pipeline is as follows:
The white balance is encoded by two values, a 4-channel white-balance
gain vector (applied in the Bayer domain), and a 3x3 color transform
matrix (applied after demosaic).
The 4-channel white-balance gains are defined as:
android.colorCorrection.gains = [ R G_even G_odd B ]
where `G_even` is the gain for green pixels on even rows of the
output, and `G_odd` is the gain for green pixels on the odd rows.
These may be identical for a given camera device implementation; if
the camera device does not support a separate gain for even/odd green
channels, it will use the `G_even` value, and write `G_odd` equal to
`G_even` in the output result metadata.
The matrices for color transforms are defined as a 9-entry vector:
android.colorCorrection.transform = [ I0 I1 I2 I3 I4 I5 I6 I7 I8 ]
which define a transform from input sensor colors, `P_in = [ r g b ]`,
to output linear sRGB, `P_out = [ r' g' b' ]`,
with colors as follows:
r' = I0r + I1g + I2b
g' = I3r + I4g + I5b
b' = I6r + I7g + I8b
Both the input and output value ranges must match. Overflow/underflow
values are clipped to fit within the range.
HAL must support both FAST and HIGH_QUALITY if color correction control is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY should generate the same output.
3
3
A color transform matrix to use to transform
from sensor RGB color space to output linear sRGB color space.
Unitless scale factors
This matrix is either set by the camera device when the request
android.colorCorrection.mode is not TRANSFORM_MATRIX, or
directly by the application in the request when the
android.colorCorrection.mode is TRANSFORM_MATRIX.
In the latter case, the camera device may round the matrix to account
for precision issues; the final rounded matrix should be reported back
in this matrix result metadata. The transform should keep the magnitude
of the output color values within `[0, 1.0]` (assuming input color
values is within the normalized range `[0, 1.0]`), or clipping may occur.
The valid range of each matrix element varies on different devices, but
values within [-1.5, 3.0] are guaranteed not to be clipped.
4
Gains applying to Bayer raw color channels for
white-balance.
Unitless gain factors
These per-channel gains are either set by the camera device
when the request android.colorCorrection.mode is not
TRANSFORM_MATRIX, or directly by the application in the
request when the android.colorCorrection.mode is
TRANSFORM_MATRIX.
The gains in the result metadata are the gains actually
applied by the camera device to the current frame.
The valid range of gains varies on different devices, but gains
between [1.0, 3.0] are guaranteed not to be clipped. Even if a given
device allows gains below 1.0, this is usually not recommended because
this can create color artifacts.
The 4-channel white-balance gains are defined in
the order of `[R G_even G_odd B]`, where `G_even` is the gain
for green pixels on even rows of the output, and `G_odd`
is the gain for green pixels on the odd rows.
If a HAL does not support a separate gain for even/odd green
channels, it must use the `G_even` value, and write
`G_odd` equal to `G_even` in the output result metadata.
OFF
No aberration correction is applied.
FAST
Aberration correction will not slow down capture rate
relative to sensor raw output.
HIGH_QUALITY
Aberration correction operates at improved quality but the capture rate might be
reduced (relative to sensor raw output rate)
Mode of operation for the chromatic aberration correction algorithm.
android.colorCorrection.availableAberrationModes
Chromatic (color) aberration is caused by the fact that different wavelengths of light
can not focus on the same point after exiting from the lens. This metadata defines
the high level control of chromatic aberration correction algorithm, which aims to
minimize the chromatic artifacts that may occur along the object boundaries in an
image.
FAST/HIGH_QUALITY both mean that camera device determined aberration
correction will be applied. HIGH_QUALITY mode indicates that the camera device will
use the highest-quality aberration correction algorithms, even if it slows down
capture rate. FAST means the camera device will not slow down capture rate when
applying aberration correction.
LEGACY devices will always be in FAST mode.
n
List of aberration correction modes for android.colorCorrection.aberrationMode that are
supported by this camera device.
Any value listed in android.colorCorrection.aberrationMode
This key lists the valid modes for android.colorCorrection.aberrationMode. If no
aberration correction modes are available for a device, this list will solely include
OFF mode. All camera devices will support either OFF or FAST mode.
Camera devices that support the MANUAL_POST_PROCESSING capability will always list
OFF mode. This includes all FULL level devices.
LEGACY devices will always only support FAST mode.
HAL must support both FAST and HIGH_QUALITY if chromatic aberration control is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY will generate the same output.
OFF
The camera device will not adjust exposure duration to
avoid banding problems.
50HZ
The camera device will adjust exposure duration to
avoid banding problems with 50Hz illumination sources.
60HZ
The camera device will adjust exposure duration to
avoid banding problems with 60Hz illumination
sources.
AUTO
The camera device will automatically adapt its
antibanding routine to the current illumination
condition. This is the default mode if AUTO is
available on given camera device.
The desired setting for the camera device's auto-exposure
algorithm's antibanding compensation.
android.control.aeAvailableAntibandingModes
Some kinds of lighting fixtures, such as some fluorescent
lights, flicker at the rate of the power supply frequency
(60Hz or 50Hz, depending on country). While this is
typically not noticeable to a person, it can be visible to
a camera device. If a camera sets its exposure time to the
wrong value, the flicker may become visible in the
viewfinder as flicker or in a final captured image, as a
set of variable-brightness bands across the image.
Therefore, the auto-exposure routines of camera devices
include antibanding routines that ensure that the chosen
exposure value will not cause such banding. The choice of
exposure time depends on the rate of flicker, which the
camera device can detect automatically, or the expected
rate can be selected by the application using this
control.
A given camera device may not support all of the possible
options for the antibanding mode. The
android.control.aeAvailableAntibandingModes key contains
the available modes for a given camera device.
AUTO mode is the default if it is available on given
camera device. When AUTO mode is not available, the
default will be either 50HZ or 60HZ, and both 50HZ
and 60HZ will be available.
If manual exposure control is enabled (by setting
android.control.aeMode or android.control.mode to OFF),
then this setting has no effect, and the application must
ensure it selects exposure times that do not cause banding
issues. The android.statistics.sceneFlicker key can assist
the application in this.
For all capture request templates, this field must be set
to AUTO if AUTO mode is available. If AUTO is not available,
the default must be either 50HZ or 60HZ, and both 50HZ and
60HZ must be available.
If manual exposure control is enabled (by setting
android.control.aeMode or android.control.mode to OFF),
then the exposure values provided by the application must not be
adjusted for antibanding.
Adjustment to auto-exposure (AE) target image
brightness.
Compensation steps
android.control.aeCompensationRange
The adjustment is measured as a count of steps, with the
step size defined by android.control.aeCompensationStep and the
allowed range by android.control.aeCompensationRange.
For example, if the exposure value (EV) step is 0.333, '6'
will mean an exposure compensation of +2 EV; -3 will mean an
exposure compensation of -1 EV. One EV represents a doubling
of image brightness. Note that this control will only be
effective if android.control.aeMode `!=` OFF. This control
will take effect even when android.control.aeLock `== true`.
In the event of exposure compensation value being changed, camera device
may take several frames to reach the newly requested exposure target.
During that time, android.control.aeState field will be in the SEARCHING
state. Once the new exposure target is reached, android.control.aeState will
change from SEARCHING to either CONVERGED, LOCKED (if AE lock is enabled), or
FLASH_REQUIRED (if the scene is too dark for still capture).
OFF
Auto-exposure lock is disabled; the AE algorithm
is free to update its parameters.
ON
Auto-exposure lock is enabled; the AE algorithm
must not update the exposure and sensitivity parameters
while the lock is active.
android.control.aeExposureCompensation setting changes
will still take effect while auto-exposure is locked.
Some rare LEGACY devices may not support
this, in which case the value will always be overridden to OFF.
Whether auto-exposure (AE) is currently locked to its latest
calculated values.
When set to `true` (ON), the AE algorithm is locked to its latest parameters,
and will not change exposure settings until the lock is set to `false` (OFF).
Note that even when AE is locked, the flash may be fired if
the android.control.aeMode is ON_AUTO_FLASH /
ON_ALWAYS_FLASH / ON_AUTO_FLASH_REDEYE.
When android.control.aeExposureCompensation is changed, even if the AE lock
is ON, the camera device will still adjust its exposure value.
If AE precapture is triggered (see android.control.aePrecaptureTrigger)
when AE is already locked, the camera device will not change the exposure time
(android.sensor.exposureTime) and sensitivity (android.sensor.sensitivity)
parameters. The flash may be fired if the android.control.aeMode
is ON_AUTO_FLASH/ON_AUTO_FLASH_REDEYE and the scene is too dark. If the
android.control.aeMode is ON_ALWAYS_FLASH, the scene may become overexposed.
Similarly, AE precapture trigger CANCEL has no effect when AE is already locked.
When an AE precapture sequence is triggered, AE unlock will not be able to unlock
the AE if AE is locked by the camera device internally during precapture metering
sequence In other words, submitting requests with AE unlock has no effect for an
ongoing precapture metering sequence. Otherwise, the precapture metering sequence
will never succeed in a sequence of preview requests where AE lock is always set
to `false`.
Since the camera device has a pipeline of in-flight requests, the settings that
get locked do not necessarily correspond to the settings that were present in the
latest capture result received from the camera device, since additional captures
and AE updates may have occurred even before the result was sent out. If an
application is switching between automatic and manual control and wishes to eliminate
any flicker during the switch, the following procedure is recommended:
1. Starting in auto-AE mode:
2. Lock AE
3. Wait for the first result to be output that has the AE locked
4. Copy exposure settings from that result into a request, set the request to manual AE
5. Submit the capture request, proceed to run manual AE as desired.
See android.control.aeState for AE lock related state transition details.
OFF
The camera device's autoexposure routine is disabled.
The application-selected android.sensor.exposureTime,
android.sensor.sensitivity and
android.sensor.frameDuration are used by the camera
device, along with android.flash.* fields, if there's
a flash unit for this camera device.
Note that auto-white balance (AWB) and auto-focus (AF)
behavior is device dependent when AE is in OFF mode.
To have consistent behavior across different devices,
it is recommended to either set AWB and AF to OFF mode
or lock AWB and AF before setting AE to OFF.
See android.control.awbMode, android.control.afMode,
android.control.awbLock, and android.control.afTrigger
for more details.
LEGACY devices do not support the OFF mode and will
override attempts to use this value to ON.
ON
The camera device's autoexposure routine is active,
with no flash control.
The application's values for
android.sensor.exposureTime,
android.sensor.sensitivity, and
android.sensor.frameDuration are ignored. The
application has control over the various
android.flash.* fields.
ON_AUTO_FLASH
Like ON, except that the camera device also controls
the camera's flash unit, firing it in low-light
conditions.
The flash may be fired during a precapture sequence
(triggered by android.control.aePrecaptureTrigger) and
may be fired for captures for which the
android.control.captureIntent field is set to
STILL_CAPTURE
ON_ALWAYS_FLASH
Like ON, except that the camera device also controls
the camera's flash unit, always firing it for still
captures.
The flash may be fired during a precapture sequence
(triggered by android.control.aePrecaptureTrigger) and
will always be fired for captures for which the
android.control.captureIntent field is set to
STILL_CAPTURE
ON_AUTO_FLASH_REDEYE
Like ON_AUTO_FLASH, but with automatic red eye
reduction.
If deemed necessary by the camera device, a red eye
reduction flash will fire during the precapture
sequence.
ON_EXTERNAL_FLASH
An external flash has been turned on.
It informs the camera device that an external flash has been turned on, and that
metering (and continuous focus if active) should be quickly recaculated to account
for the external flash. Otherwise, this mode acts like ON.
When the external flash is turned off, AE mode should be changed to one of the
other available AE modes.
If the camera device supports AE external flash mode, android.control.aeState must
be FLASH_REQUIRED after the camera device finishes AE scan and it's too dark without
flash.
The desired mode for the camera device's
auto-exposure routine.
android.control.aeAvailableModes
This control is only effective if android.control.mode is
AUTO.
When set to any of the ON modes, the camera device's
auto-exposure routine is enabled, overriding the
application's selected exposure time, sensor sensitivity,
and frame duration (android.sensor.exposureTime,
android.sensor.sensitivity, and
android.sensor.frameDuration). If one of the FLASH modes
is selected, the camera device's flash unit controls are
also overridden.
The FLASH modes are only available if the camera device
has a flash unit (android.flash.info.available is `true`).
If flash TORCH mode is desired, this field must be set to
ON or OFF, and android.flash.mode set to TORCH.
When set to any of the ON modes, the values chosen by the
camera device auto-exposure routine for the overridden
fields for a given capture will be available in its
CaptureResult.
5
area_count
List of metering areas to use for auto-exposure adjustment.
Pixel coordinates within android.sensor.info.activeArraySize or
android.sensor.info.preCorrectionActiveArraySize depending on
distortion correction capability and mode
Coordinates must be between `[(0,0), (width, height))` of
android.sensor.info.activeArraySize or android.sensor.info.preCorrectionActiveArraySize
depending on distortion correction capability and mode
Not available if android.control.maxRegionsAe is 0.
Otherwise will always be present.
The maximum number of regions supported by the device is determined by the value
of android.control.maxRegionsAe.
For devices not supporting android.distortionCorrection.mode control, the coordinate
system always follows that of android.sensor.info.activeArraySize, with (0,0) being
the top-left pixel in the active pixel array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the
active pixel array.
For devices supporting android.distortionCorrection.mode control, the coordinate
system depends on the mode being set.
When the distortion correction mode is OFF, the coordinate system follows
android.sensor.info.preCorrectionActiveArraySize, with
`(0, 0)` being the top-left pixel of the pre-correction active array, and
(android.sensor.info.preCorrectionActiveArraySize.width - 1,
android.sensor.info.preCorrectionActiveArraySize.height - 1) being the bottom-right
pixel in the pre-correction active pixel array.
When the distortion correction mode is not OFF, the coordinate system follows
android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the
active pixel array.
The weight must be within `[0, 1000]`, and represents a weight
for every pixel in the area. This means that a large metering area
with the same weight as a smaller area will have more effect in
the metering result. Metering areas can partially overlap and the
camera device will add the weights in the overlap region.
The weights are relative to weights of other exposure metering regions, so if only one
region is used, all non-zero weights will have the same effect. A region with 0
weight is ignored.
If all regions have 0 weight, then no specific metering area needs to be used by the
camera device.
If the metering region is outside the used android.scaler.cropRegion returned in
capture result metadata, the camera device will ignore the sections outside the crop
region and output only the intersection rectangle as the metering region in the result
metadata. If the region is entirely outside the crop region, it will be ignored and
not reported in the result metadata.
The data representation is `int[5 * area_count]`.
Every five elements represent a metering region of `(xmin, ymin, xmax, ymax, weight)`.
The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and
ymax.
The HAL level representation of MeteringRectangle[] is a
int[5 * area_count].
Every five elements represent a metering region of
(xmin, ymin, xmax, ymax, weight).
The rectangle is defined to be inclusive on xmin and ymin, but
exclusive on xmax and ymax.
HAL must always report metering regions in the coordinate system of pre-correction
active array.
2
Range over which the auto-exposure routine can
adjust the capture frame rate to maintain good
exposure.
Frames per second (FPS)
Any of the entries in android.control.aeAvailableTargetFpsRanges
Only constrains auto-exposure (AE) algorithm, not
manual control of android.sensor.exposureTime and
android.sensor.frameDuration.
IDLE
The trigger is idle.
START
The precapture metering sequence will be started
by the camera device.
The exact effect of the precapture trigger depends on
the current AE mode and state.
CANCEL
The camera device will cancel any currently active or completed
precapture metering sequence, the auto-exposure routine will return to its
initial state.
Whether the camera device will trigger a precapture
metering sequence when it processes this request.
This entry is normally set to IDLE, or is not
included at all in the request settings. When included and
set to START, the camera device will trigger the auto-exposure (AE)
precapture metering sequence.
When set to CANCEL, the camera device will cancel any active
precapture metering trigger, and return to its initial AE state.
If a precapture metering sequence is already completed, and the camera
device has implicitly locked the AE for subsequent still capture, the
CANCEL trigger will unlock the AE and return to its initial AE state.
The precapture sequence should be triggered before starting a
high-quality still capture for final metering decisions to
be made, and for firing pre-capture flash pulses to estimate
scene brightness and required final capture flash power, when
the flash is enabled.
Normally, this entry should be set to START for only a
single request, and the application should wait until the
sequence completes before starting a new one.
When a precapture metering sequence is finished, the camera device
may lock the auto-exposure routine internally to be able to accurately expose the
subsequent still capture image (`android.control.captureIntent == STILL_CAPTURE`).
For this case, the AE may not resume normal scan if no subsequent still capture is
submitted. To ensure that the AE routine restarts normal scan, the application should
submit a request with `android.control.aeLock == true`, followed by a request
with `android.control.aeLock == false`, if the application decides not to submit a
still capture request after the precapture sequence completes. Alternatively, for
API level 23 or newer devices, the CANCEL can be used to unlock the camera device
internally locked AE if the application doesn't submit a still capture request after
the AE precapture trigger. Note that, the CANCEL was added in API level 23, and must not
be used in devices that have earlier API levels.
The exact effect of auto-exposure (AE) precapture trigger
depends on the current AE mode and state; see
android.control.aeState for AE precapture state transition
details.
On LEGACY-level devices, the precapture trigger is not supported;
capturing a high-resolution JPEG image will automatically trigger a
precapture sequence before the high-resolution capture, including
potentially firing a pre-capture flash.
Using the precapture trigger and the auto-focus trigger android.control.afTrigger
simultaneously is allowed. However, since these triggers often require cooperation between
the auto-focus and auto-exposure routines (for example, the may need to be enabled for a
focus sweep), the camera device may delay acting on a later trigger until the previous
trigger has been fully handled. This may lead to longer intervals between the trigger and
changes to android.control.aeState indicating the start of the precapture sequence, for
example.
If both the precapture and the auto-focus trigger are activated on the same request, then
the camera device will complete them in the optimal order for that device.
The HAL must support triggering the AE precapture trigger while an AF trigger is active
(and vice versa), or at the same time as the AF trigger. It is acceptable for the HAL to
treat these as two consecutive triggers, for example handling the AF trigger and then the
AE trigger. Or the HAL may choose to optimize the case with both triggers fired at once,
to minimize the latency for converging both focus and exposure/flash usage.
OFF
The auto-focus routine does not control the lens;
android.lens.focusDistance is controlled by the
application.
AUTO
Basic automatic focus mode.
In this mode, the lens does not move unless
the autofocus trigger action is called. When that trigger
is activated, AF will transition to ACTIVE_SCAN, then to
the outcome of the scan (FOCUSED or NOT_FOCUSED).
Always supported if lens is not fixed focus.
Use android.lens.info.minimumFocusDistance to determine if lens
is fixed-focus.
Triggering AF_CANCEL resets the lens position to default,
and sets the AF state to INACTIVE.
MACRO
Close-up focusing mode.
In this mode, the lens does not move unless the
autofocus trigger action is called. When that trigger is
activated, AF will transition to ACTIVE_SCAN, then to
the outcome of the scan (FOCUSED or NOT_FOCUSED). This
mode is optimized for focusing on objects very close to
the camera.
When that trigger is activated, AF will transition to
ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or
NOT_FOCUSED). Triggering cancel AF resets the lens
position to default, and sets the AF state to
INACTIVE.
CONTINUOUS_VIDEO
In this mode, the AF algorithm modifies the lens
position continually to attempt to provide a
constantly-in-focus image stream.
The focusing behavior should be suitable for good quality
video recording; typically this means slower focus
movement and no overshoots. When the AF trigger is not
involved, the AF algorithm should start in INACTIVE state,
and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED
states as appropriate. When the AF trigger is activated,
the algorithm should immediately transition into
AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the
lens position until a cancel AF trigger is received.
Once cancel is received, the algorithm should transition
back to INACTIVE and resume passive scan. Note that this
behavior is not identical to CONTINUOUS_PICTURE, since an
ongoing PASSIVE_SCAN must immediately be
canceled.
CONTINUOUS_PICTURE
In this mode, the AF algorithm modifies the lens
position continually to attempt to provide a
constantly-in-focus image stream.
The focusing behavior should be suitable for still image
capture; typically this means focusing as fast as
possible. When the AF trigger is not involved, the AF
algorithm should start in INACTIVE state, and then
transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as
appropriate as it attempts to maintain focus. When the AF
trigger is activated, the algorithm should finish its
PASSIVE_SCAN if active, and then transition into
AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the
lens position until a cancel AF trigger is received.
When the AF cancel trigger is activated, the algorithm
should transition back to INACTIVE and then act as if it
has just been started.
EDOF
Extended depth of field (digital focus) mode.
The camera device will produce images with an extended
depth of field automatically; no special focusing
operations need to be done before taking a picture.
AF triggers are ignored, and the AF state will always be
INACTIVE.
Whether auto-focus (AF) is currently enabled, and what
mode it is set to.
android.control.afAvailableModes
Only effective if android.control.mode = AUTO and the lens is not fixed focus
(i.e. `android.lens.info.minimumFocusDistance > 0`). Also note that
when android.control.aeMode is OFF, the behavior of AF is device
dependent. It is recommended to lock AF by using android.control.afTrigger before
setting android.control.aeMode to OFF, or set AF mode to OFF when AE is OFF.
If the lens is controlled by the camera device auto-focus algorithm,
the camera device will report the current AF status in android.control.afState
in result metadata.
When afMode is AUTO or MACRO, the lens must not move until an AF trigger is sent in a
request (android.control.afTrigger `==` START). After an AF trigger, the afState will end
up with either FOCUSED_LOCKED or NOT_FOCUSED_LOCKED state (see
android.control.afState for detailed state transitions), which indicates that the lens is
locked and will not move. If camera movement (e.g. tilting camera) causes the lens to move
after the lens is locked, the HAL must compensate this movement appropriately such that
the same focal plane remains in focus.
When afMode is one of the continuous auto focus modes, the HAL is free to start a AF
scan whenever it's not locked. When the lens is locked after an AF trigger
(see android.control.afState for detailed state transitions), the HAL should maintain the
same lock behavior as above.
When afMode is OFF, the application controls focus manually. The accuracy of the
focus distance control depends on the android.lens.info.focusDistanceCalibration.
However, the lens must not move regardless of the camera movement for any focus distance
manual control.
To put this in concrete terms, if the camera has lens elements which may move based on
camera orientation or motion (e.g. due to gravity), then the HAL must drive the lens to
remain in a fixed position invariant to the camera's orientation or motion, for example,
by using accelerometer measurements in the lens control logic. This is a typical issue
that will arise on camera modules with open-loop VCMs.
5
area_count
List of metering areas to use for auto-focus.
Pixel coordinates within android.sensor.info.activeArraySize or
android.sensor.info.preCorrectionActiveArraySize depending on
distortion correction capability and mode
Coordinates must be between `[(0,0), (width, height))` of
android.sensor.info.activeArraySize or android.sensor.info.preCorrectionActiveArraySize
depending on distortion correction capability and mode
Not available if android.control.maxRegionsAf is 0.
Otherwise will always be present.
The maximum number of focus areas supported by the device is determined by the value
of android.control.maxRegionsAf.
For devices not supporting android.distortionCorrection.mode control, the coordinate
system always follows that of android.sensor.info.activeArraySize, with (0,0) being
the top-left pixel in the active pixel array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the
active pixel array.
For devices supporting android.distortionCorrection.mode control, the coordinate
system depends on the mode being set.
When the distortion correction mode is OFF, the coordinate system follows
android.sensor.info.preCorrectionActiveArraySize, with
`(0, 0)` being the top-left pixel of the pre-correction active array, and
(android.sensor.info.preCorrectionActiveArraySize.width - 1,
android.sensor.info.preCorrectionActiveArraySize.height - 1) being the bottom-right
pixel in the pre-correction active pixel array.
When the distortion correction mode is not OFF, the coordinate system follows
android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the
active pixel array.
The weight must be within `[0, 1000]`, and represents a weight
for every pixel in the area. This means that a large metering area
with the same weight as a smaller area will have more effect in
the metering result. Metering areas can partially overlap and the
camera device will add the weights in the overlap region.
The weights are relative to weights of other metering regions, so if only one region
is used, all non-zero weights will have the same effect. A region with 0 weight is
ignored.
If all regions have 0 weight, then no specific metering area needs to be used by the
camera device. The capture result will either be a zero weight region as well, or
the region selected by the camera device as the focus area of interest.
If the metering region is outside the used android.scaler.cropRegion returned in
capture result metadata, the camera device will ignore the sections outside the crop
region and output only the intersection rectangle as the metering region in the result
metadata. If the region is entirely outside the crop region, it will be ignored and
not reported in the result metadata.
The data representation is `int[5 * area_count]`.
Every five elements represent a metering region of `(xmin, ymin, xmax, ymax, weight)`.
The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and
ymax.
The HAL level representation of MeteringRectangle[] is a
int[5 * area_count].
Every five elements represent a metering region of
(xmin, ymin, xmax, ymax, weight).
The rectangle is defined to be inclusive on xmin and ymin, but
exclusive on xmax and ymax.
HAL must always report metering regions in the coordinate system of pre-correction
active array.
IDLE
The trigger is idle.
START
Autofocus will trigger now.
CANCEL
Autofocus will return to its initial
state, and cancel any currently active trigger.
Whether the camera device will trigger autofocus for this request.
This entry is normally set to IDLE, or is not
included at all in the request settings.
When included and set to START, the camera device will trigger the
autofocus algorithm. If autofocus is disabled, this trigger has no effect.
When set to CANCEL, the camera device will cancel any active trigger,
and return to its initial AF state.
Generally, applications should set this entry to START or CANCEL for only a
single capture, and then return it to IDLE (or not set at all). Specifying
START for multiple captures in a row means restarting the AF operation over
and over again.
See android.control.afState for what the trigger means for each AF mode.
Using the autofocus trigger and the precapture trigger android.control.aePrecaptureTrigger
simultaneously is allowed. However, since these triggers often require cooperation between
the auto-focus and auto-exposure routines (for example, the may need to be enabled for a
focus sweep), the camera device may delay acting on a later trigger until the previous
trigger has been fully handled. This may lead to longer intervals between the trigger and
changes to android.control.afState, for example.
The HAL must support triggering the AF trigger while an AE precapture trigger is active
(and vice versa), or at the same time as the AE trigger. It is acceptable for the HAL to
treat these as two consecutive triggers, for example handling the AF trigger and then the
AE trigger. Or the HAL may choose to optimize the case with both triggers fired at once,
to minimize the latency for converging both focus and exposure/flash usage.
OFF
Auto-white balance lock is disabled; the AWB
algorithm is free to update its parameters if in AUTO
mode.
ON
Auto-white balance lock is enabled; the AWB
algorithm will not update its parameters while the lock
is active.
Whether auto-white balance (AWB) is currently locked to its
latest calculated values.
When set to `true` (ON), the AWB algorithm is locked to its latest parameters,
and will not change color balance settings until the lock is set to `false` (OFF).
Since the camera device has a pipeline of in-flight requests, the settings that
get locked do not necessarily correspond to the settings that were present in the
latest capture result received from the camera device, since additional captures
and AWB updates may have occurred even before the result was sent out. If an
application is switching between automatic and manual control and wishes to eliminate
any flicker during the switch, the following procedure is recommended:
1. Starting in auto-AWB mode:
2. Lock AWB
3. Wait for the first result to be output that has the AWB locked
4. Copy AWB settings from that result into a request, set the request to manual AWB
5. Submit the capture request, proceed to run manual AWB as desired.
Note that AWB lock is only meaningful when
android.control.awbMode is in the AUTO mode; in other modes,
AWB is already fixed to a specific setting.
Some LEGACY devices may not support ON; the value is then overridden to OFF.
OFF
The camera device's auto-white balance routine is disabled.
The application-selected color transform matrix
(android.colorCorrection.transform) and gains
(android.colorCorrection.gains) are used by the camera
device for manual white balance control.
AUTO
The camera device's auto-white balance routine is active.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
INCANDESCENT
The camera device's auto-white balance routine is disabled;
the camera device uses incandescent light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant A.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
FLUORESCENT
The camera device's auto-white balance routine is disabled;
the camera device uses fluorescent light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant F2.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
WARM_FLUORESCENT
The camera device's auto-white balance routine is disabled;
the camera device uses warm fluorescent light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant F4.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
DAYLIGHT
The camera device's auto-white balance routine is disabled;
the camera device uses daylight light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant D65.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
CLOUDY_DAYLIGHT
The camera device's auto-white balance routine is disabled;
the camera device uses cloudy daylight light as the assumed scene
illumination for white balance.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
TWILIGHT
The camera device's auto-white balance routine is disabled;
the camera device uses twilight light as the assumed scene
illumination for white balance.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
SHADE
The camera device's auto-white balance routine is disabled;
the camera device uses shade light as the assumed scene
illumination for white balance.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
Whether auto-white balance (AWB) is currently setting the color
transform fields, and what its illumination target
is.
android.control.awbAvailableModes
This control is only effective if android.control.mode is AUTO.
When set to the ON mode, the camera device's auto-white balance
routine is enabled, overriding the application's selected
android.colorCorrection.transform, android.colorCorrection.gains and
android.colorCorrection.mode. Note that when android.control.aeMode
is OFF, the behavior of AWB is device dependent. It is recommened to
also set AWB mode to OFF or lock AWB by using android.control.awbLock before
setting AE mode to OFF.
When set to the OFF mode, the camera device's auto-white balance
routine is disabled. The application manually controls the white
balance by android.colorCorrection.transform, android.colorCorrection.gains
and android.colorCorrection.mode.
When set to any other modes, the camera device's auto-white
balance routine is disabled. The camera device uses each
particular illumination target for white balance
adjustment. The application's values for
android.colorCorrection.transform,
android.colorCorrection.gains and
android.colorCorrection.mode are ignored.
5
area_count
List of metering areas to use for auto-white-balance illuminant
estimation.
Pixel coordinates within android.sensor.info.activeArraySize or
android.sensor.info.preCorrectionActiveArraySize depending on
distortion correction capability and mode
Coordinates must be between `[(0,0), (width, height))` of
android.sensor.info.activeArraySize or android.sensor.info.preCorrectionActiveArraySize
depending on distortion correction capability and mode
Not available if android.control.maxRegionsAwb is 0.
Otherwise will always be present.
The maximum number of regions supported by the device is determined by the value
of android.control.maxRegionsAwb.
For devices not supporting android.distortionCorrection.mode control, the coordinate
system always follows that of android.sensor.info.activeArraySize, with (0,0) being
the top-left pixel in the active pixel array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the
active pixel array.
For devices supporting android.distortionCorrection.mode control, the coordinate
system depends on the mode being set.
When the distortion correction mode is OFF, the coordinate system follows
android.sensor.info.preCorrectionActiveArraySize, with
`(0, 0)` being the top-left pixel of the pre-correction active array, and
(android.sensor.info.preCorrectionActiveArraySize.width - 1,
android.sensor.info.preCorrectionActiveArraySize.height - 1) being the bottom-right
pixel in the pre-correction active pixel array.
When the distortion correction mode is not OFF, the coordinate system follows
android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the
active pixel array.
The weight must range from 0 to 1000, and represents a weight
for every pixel in the area. This means that a large metering area
with the same weight as a smaller area will have more effect in
the metering result. Metering areas can partially overlap and the
camera device will add the weights in the overlap region.
The weights are relative to weights of other white balance metering regions, so if
only one region is used, all non-zero weights will have the same effect. A region with
0 weight is ignored.
If all regions have 0 weight, then no specific metering area needs to be used by the
camera device.
If the metering region is outside the used android.scaler.cropRegion returned in
capture result metadata, the camera device will ignore the sections outside the crop
region and output only the intersection rectangle as the metering region in the result
metadata. If the region is entirely outside the crop region, it will be ignored and
not reported in the result metadata.
The data representation is `int[5 * area_count]`.
Every five elements represent a metering region of `(xmin, ymin, xmax, ymax, weight)`.
The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and
ymax.
The HAL level representation of MeteringRectangle[] is a
int[5 * area_count].
Every five elements represent a metering region of
(xmin, ymin, xmax, ymax, weight).
The rectangle is defined to be inclusive on xmin and ymin, but
exclusive on xmax and ymax.
HAL must always report metering regions in the coordinate system of pre-correction
active array.
CUSTOM
The goal of this request doesn't fall into the other
categories. The camera device will default to preview-like
behavior.
PREVIEW
This request is for a preview-like use case.
The precapture trigger may be used to start off a metering
w/flash sequence.
STILL_CAPTURE
This request is for a still capture-type
use case.
If the flash unit is under automatic control, it may fire as needed.
VIDEO_RECORD
This request is for a video recording
use case.
VIDEO_SNAPSHOT
This request is for a video snapshot (still
image while recording video) use case.
The camera device should take the highest-quality image
possible (given the other settings) without disrupting the
frame rate of video recording.
ZERO_SHUTTER_LAG
This request is for a ZSL usecase; the
application will stream full-resolution images and
reprocess one or several later for a final
capture.
MANUAL
This request is for manual capture use case where
the applications want to directly control the capture parameters.
For example, the application may wish to manually control
android.sensor.exposureTime, android.sensor.sensitivity, etc.
MOTION_TRACKING
This request is for a motion tracking use case, where
the application will use camera and inertial sensor data to
locate and track objects in the world.
The camera device auto-exposure routine will limit the exposure time
of the camera to no more than 20 milliseconds, to minimize motion blur.
Information to the camera device 3A (auto-exposure,
auto-focus, auto-white balance) routines about the purpose
of this capture, to help the camera device to decide optimal 3A
strategy.
This control (except for MANUAL) is only effective if
`android.control.mode != OFF` and any 3A routine is active.
All intents are supported by all devices, except that:
* ZERO_SHUTTER_LAG will be supported if android.request.availableCapabilities contains
PRIVATE_REPROCESSING or YUV_REPROCESSING.
* MANUAL will be supported if android.request.availableCapabilities contains
MANUAL_SENSOR.
* MOTION_TRACKING will be supported if android.request.availableCapabilities contains
MOTION_TRACKING.
OFF
No color effect will be applied.
MONO
A "monocolor" effect where the image is mapped into
a single color.
This will typically be grayscale.
NEGATIVE
A "photo-negative" effect where the image's colors
are inverted.
SOLARIZE
A "solarisation" effect (Sabattier effect) where the
image is wholly or partially reversed in
tone.
SEPIA
A "sepia" effect where the image is mapped into warm
gray, red, and brown tones.
POSTERIZE
A "posterization" effect where the image uses
discrete regions of tone rather than a continuous
gradient of tones.
WHITEBOARD
A "whiteboard" effect where the image is typically displayed
as regions of white, with black or grey details.
BLACKBOARD
A "blackboard" effect where the image is typically displayed
as regions of black, with white or grey details.
AQUA
An "aqua" effect where a blue hue is added to the image.
A special color effect to apply.
android.control.availableEffects
When this mode is set, a color effect will be applied
to images produced by the camera device. The interpretation
and implementation of these color effects is left to the
implementor of the camera device, and should not be
depended on to be consistent (or present) across all
devices.
OFF
Full application control of pipeline.
All control by the device's metering and focusing (3A)
routines is disabled, and no other settings in
android.control.* have any effect, except that
android.control.captureIntent may be used by the camera
device to select post-processing values for processing
blocks that do not allow for manual control, or are not
exposed by the camera API.
However, the camera device's 3A routines may continue to
collect statistics and update their internal state so that
when control is switched to AUTO mode, good control values
can be immediately applied.
AUTO
Use settings for each individual 3A routine.
Manual control of capture parameters is disabled. All
controls in android.control.* besides sceneMode take
effect.
USE_SCENE_MODE
Use a specific scene mode.
Enabling this disables control.aeMode, control.awbMode and
control.afMode controls; the camera device will ignore
those settings while USE_SCENE_MODE is active (except for
FACE_PRIORITY scene mode). Other control entries are still active.
This setting can only be used if scene mode is supported (i.e.
android.control.availableSceneModes
contain some modes other than DISABLED).
OFF_KEEP_STATE
Same as OFF mode, except that this capture will not be
used by camera device background auto-exposure, auto-white balance and
auto-focus algorithms (3A) to update their statistics.
Specifically, the 3A routines are locked to the last
values set from a request with AUTO, OFF, or
USE_SCENE_MODE, and any statistics or state updates
collected from manual captures with OFF_KEEP_STATE will be
discarded by the camera device.
Overall mode of 3A (auto-exposure, auto-white-balance, auto-focus) control
routines.
android.control.availableModes
This is a top-level 3A control switch. When set to OFF, all 3A control
by the camera device is disabled. The application must set the fields for
capture parameters itself.
When set to AUTO, the individual algorithm controls in
android.control.* are in effect, such as android.control.afMode.
When set to USE_SCENE_MODE, the individual controls in
android.control.* are mostly disabled, and the camera device
implements one of the scene mode settings (such as ACTION,
SUNSET, or PARTY) as it wishes. The camera device scene mode
3A settings are provided by {@link
android.hardware.camera2.CaptureResult|ACameraCaptureSession_captureCallback_result
capture results}.
When set to OFF_KEEP_STATE, it is similar to OFF mode, the only difference
is that this frame will not be used by camera device background 3A statistics
update, as if this frame is never captured. This mode can be used in the scenario
where the application doesn't want a 3A manual control capture to affect
the subsequent auto 3A capture results.
DISABLED
Indicates that no scene modes are set for a given capture request.
FACE_PRIORITY
If face detection support exists, use face
detection data for auto-focus, auto-white balance, and
auto-exposure routines.
If face detection statistics are disabled
(i.e. android.statistics.faceDetectMode is set to OFF),
this should still operate correctly (but will not return
face detection statistics to the framework).
Unlike the other scene modes, android.control.aeMode,
android.control.awbMode, and android.control.afMode
remain active when FACE_PRIORITY is set.
ACTION
Optimized for photos of quickly moving objects.
Similar to SPORTS.
PORTRAIT
Optimized for still photos of people.
LANDSCAPE
Optimized for photos of distant macroscopic objects.
NIGHT
Optimized for low-light settings.
NIGHT_PORTRAIT
Optimized for still photos of people in low-light
settings.
THEATRE
Optimized for dim, indoor settings where flash must
remain off.
BEACH
Optimized for bright, outdoor beach settings.
SNOW
Optimized for bright, outdoor settings containing snow.
SUNSET
Optimized for scenes of the setting sun.
STEADYPHOTO
Optimized to avoid blurry photos due to small amounts of
device motion (for example: due to hand shake).
FIREWORKS
Optimized for nighttime photos of fireworks.
SPORTS
Optimized for photos of quickly moving people.
Similar to ACTION.
PARTY
Optimized for dim, indoor settings with multiple moving
people.
CANDLELIGHT
Optimized for dim settings where the main light source
is a candle.
BARCODE
Optimized for accurately capturing a photo of barcode
for use by camera applications that wish to read the
barcode value.
HIGH_SPEED_VIDEO
This is deprecated, please use {@link
android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}
and {@link
android.hardware.camera2.CameraConstrainedHighSpeedCaptureSession#createHighSpeedRequestList}
for high speed video recording.
Optimized for high speed video recording (frame rate >=60fps) use case.
The supported high speed video sizes and fps ranges are specified in
android.control.availableHighSpeedVideoConfigurations. To get desired
output frame rates, the application is only allowed to select video size
and fps range combinations listed in this static metadata. The fps range
can be control via android.control.aeTargetFpsRange.
In this mode, the camera device will override aeMode, awbMode, and afMode to
ON, ON, and CONTINUOUS_VIDEO, respectively. All post-processing block mode
controls will be overridden to be FAST. Therefore, no manual control of capture
and post-processing parameters is possible. All other controls operate the
same as when android.control.mode == AUTO. This means that all other
android.control.* fields continue to work, such as
* android.control.aeTargetFpsRange
* android.control.aeExposureCompensation
* android.control.aeLock
* android.control.awbLock
* android.control.effectMode
* android.control.aeRegions
* android.control.afRegions
* android.control.awbRegions
* android.control.afTrigger
* android.control.aePrecaptureTrigger
Outside of android.control.*, the following controls will work:
* android.flash.mode (automatic flash for still capture will not work since aeMode is ON)
* android.lens.opticalStabilizationMode (if it is supported)
* android.scaler.cropRegion
* android.statistics.faceDetectMode
For high speed recording use case, the actual maximum supported frame rate may
be lower than what camera can output, depending on the destination Surfaces for
the image data. For example, if the destination surface is from video encoder,
the application need check if the video encoder is capable of supporting the
high frame rate for a given video size, or it will end up with lower recording
frame rate. If the destination surface is from preview window, the preview frame
rate will be bounded by the screen refresh rate.
The camera device will only support up to 2 output high speed streams
(processed non-stalling format defined in android.request.maxNumOutputStreams)
in this mode. This control will be effective only if all of below conditions are true:
* The application created no more than maxNumHighSpeedStreams processed non-stalling
format output streams, where maxNumHighSpeedStreams is calculated as
min(2, android.request.maxNumOutputStreams[Processed (but not-stalling)]).
* The stream sizes are selected from the sizes reported by
android.control.availableHighSpeedVideoConfigurations.
* No processed non-stalling or raw streams are configured.
When above conditions are NOT satistied, the controls of this mode and
android.control.aeTargetFpsRange will be ignored by the camera device,
the camera device will fall back to android.control.mode `==` AUTO,
and the returned capture result metadata will give the fps range choosen
by the camera device.
Switching into or out of this mode may trigger some camera ISP/sensor
reconfigurations, which may introduce extra latency. It is recommended that
the application avoids unnecessary scene mode switch as much as possible.
HDR
Turn on a device-specific high dynamic range (HDR) mode.
In this scene mode, the camera device captures images
that keep a larger range of scene illumination levels
visible in the final image. For example, when taking a
picture of a object in front of a bright window, both
the object and the scene through the window may be
visible when using HDR mode, while in normal AUTO mode,
one or the other may be poorly exposed. As a tradeoff,
HDR mode generally takes much longer to capture a single
image, has no user control, and may have other artifacts
depending on the HDR method used.
Therefore, HDR captures operate at a much slower rate
than regular captures.
In this mode, on LIMITED or FULL devices, when a request
is made with a android.control.captureIntent of
STILL_CAPTURE, the camera device will capture an image
using a high dynamic range capture technique. On LEGACY
devices, captures that target a JPEG-format output will
be captured with HDR, and the capture intent is not
relevant.
The HDR capture may involve the device capturing a burst
of images internally and combining them into one, or it
may involve the device using specialized high dynamic
range capture hardware. In all cases, a single image is
produced in response to a capture request submitted
while in HDR mode.
Since substantial post-processing is generally needed to
produce an HDR image, only YUV, PRIVATE, and JPEG
outputs are supported for LIMITED/FULL device HDR
captures, and only JPEG outputs are supported for LEGACY
HDR captures. Using a RAW output for HDR capture is not
supported.
Some devices may also support always-on HDR, which
applies HDR processing at full frame rate. For these
devices, intents other than STILL_CAPTURE will also
produce an HDR output with no frame rate impact compared
to normal operation, though the quality may be lower
than for STILL_CAPTURE intents.
If SCENE_MODE_HDR is used with unsupported output types
or capture intents, the images captured will be as if
the SCENE_MODE was not enabled at all.
FACE_PRIORITY_LOW_LIGHT
Same as FACE_PRIORITY scene mode, except that the camera
device will choose higher sensitivity values (android.sensor.sensitivity)
under low light conditions.
The camera device may be tuned to expose the images in a reduced
sensitivity range to produce the best quality images. For example,
if the android.sensor.info.sensitivityRange gives range of [100, 1600],
the camera device auto-exposure routine tuning process may limit the actual
exposure sensitivity range to [100, 1200] to ensure that the noise level isn't
exessive in order to preserve the image quality. Under this situation, the image under
low light may be under-exposed when the sensor max exposure time (bounded by the
android.control.aeTargetFpsRange when android.control.aeMode is one of the
ON_* modes) and effective max sensitivity are reached. This scene mode allows the
camera device auto-exposure routine to increase the sensitivity up to the max
sensitivity specified by android.sensor.info.sensitivityRange when the scene is too
dark and the max exposure time is reached. The captured images may be noisier
compared with the images captured in normal FACE_PRIORITY mode; therefore, it is
recommended that the application only use this scene mode when it is capable of
reducing the noise level of the captured images.
Unlike the other scene modes, android.control.aeMode,
android.control.awbMode, and android.control.afMode
remain active when FACE_PRIORITY_LOW_LIGHT is set.
DEVICE_CUSTOM_START
Scene mode values within the range of
`[DEVICE_CUSTOM_START, DEVICE_CUSTOM_END]` are reserved for device specific
customized scene modes.
DEVICE_CUSTOM_END
Scene mode values within the range of
`[DEVICE_CUSTOM_START, DEVICE_CUSTOM_END]` are reserved for device specific
customized scene modes.
Control for which scene mode is currently active.
android.control.availableSceneModes
Scene modes are custom camera modes optimized for a certain set of conditions and
capture settings.
This is the mode that that is active when
`android.control.mode == USE_SCENE_MODE`. Aside from FACE_PRIORITY, these modes will
disable android.control.aeMode, android.control.awbMode, and android.control.afMode
while in use.
The interpretation and implementation of these scene modes is left
to the implementor of the camera device. Their behavior will not be
consistent across all devices, and any given device may only implement
a subset of these modes.
HAL implementations that include scene modes are expected to provide
the per-scene settings to use for android.control.aeMode,
android.control.awbMode, and android.control.afMode in
android.control.sceneModeOverrides.
For HIGH_SPEED_VIDEO mode, if it is included in android.control.availableSceneModes, the
HAL must list supported video size and fps range in
android.control.availableHighSpeedVideoConfigurations. For a given size, e.g. 1280x720,
if the HAL has two different sensor configurations for normal streaming mode and high
speed streaming, when this scene mode is set/reset in a sequence of capture requests, the
HAL may have to switch between different sensor modes. This mode is deprecated in legacy
HAL3.3, to support high speed video recording, please implement
android.control.availableHighSpeedVideoConfigurations and CONSTRAINED_HIGH_SPEED_VIDEO
capbility defined in android.request.availableCapabilities.
OFF
Video stabilization is disabled.
ON
Video stabilization is enabled.
Whether video stabilization is
active.
Video stabilization automatically warps images from
the camera in order to stabilize motion between consecutive frames.
If enabled, video stabilization can modify the
android.scaler.cropRegion to keep the video stream stabilized.
Switching between different video stabilization modes may take several
frames to initialize, the camera device will report the current mode
in capture result metadata. For example, When "ON" mode is requested,
the video stabilization modes in the first several capture results may
still be "OFF", and it will become "ON" when the initialization is
done.
In addition, not all recording sizes or frame rates may be supported for
stabilization by a device that reports stabilization support. It is guaranteed
that an output targeting a MediaRecorder or MediaCodec will be stabilized if
the recording resolution is less than or equal to 1920 x 1080 (width less than
or equal to 1920, height less than or equal to 1080), and the recording
frame rate is less than or equal to 30fps. At other sizes, the CaptureResult
android.control.videoStabilizationMode field will return
OFF if the recording output is not stabilized, or if there are no output
Surface types that can be stabilized.
If a camera device supports both this mode and OIS
(android.lens.opticalStabilizationMode), turning both modes on may
produce undesirable interaction, so it is recommended not to enable
both at the same time.
n
List of auto-exposure antibanding modes for android.control.aeAntibandingMode that are
supported by this camera device.
Any value listed in android.control.aeAntibandingMode
Not all of the auto-exposure anti-banding modes may be
supported by a given camera device. This field lists the
valid anti-banding modes that the application may request
for this camera device with the
android.control.aeAntibandingMode control.
n
List of auto-exposure modes for android.control.aeMode that are supported by this camera
device.
Any value listed in android.control.aeMode
Not all the auto-exposure modes may be supported by a
given camera device, especially if no flash unit is
available. This entry lists the valid modes for
android.control.aeMode for this camera device.
All camera devices support ON, and all camera devices with flash
units support ON_AUTO_FLASH and ON_ALWAYS_FLASH.
FULL mode camera devices always support OFF mode,
which enables application control of camera exposure time,
sensitivity, and frame duration.
LEGACY mode camera devices never support OFF mode.
LIMITED mode devices support OFF if they support the MANUAL_SENSOR
capability.
2
n
List of frame rate ranges for android.control.aeTargetFpsRange supported by
this camera device.
Frames per second (FPS)
For devices at the LEGACY level or above:
* For constant-framerate recording, for each normal
{@link android.media.CamcorderProfile CamcorderProfile}, that is, a
{@link android.media.CamcorderProfile CamcorderProfile} that has
{@link android.media.CamcorderProfile#quality quality} in
the range [{@link android.media.CamcorderProfile#QUALITY_LOW QUALITY_LOW},
{@link android.media.CamcorderProfile#QUALITY_2160P QUALITY_2160P}], if the profile is
supported by the device and has
{@link android.media.CamcorderProfile#videoFrameRate videoFrameRate} `x`, this list will
always include (`x`,`x`).
* Also, a camera device must either not support any
{@link android.media.CamcorderProfile CamcorderProfile},
or support at least one
normal {@link android.media.CamcorderProfile CamcorderProfile} that has
{@link android.media.CamcorderProfile#videoFrameRate videoFrameRate} `x` >= 24.
For devices at the LIMITED level or above:
* For YUV_420_888 burst capture use case, this list will always include (`min`, `max`)
and (`max`, `max`) where `min` <= 15 and `max` = the maximum output frame rate of the
maximum YUV_420_888 output size.
2
Maximum and minimum exposure compensation values for
android.control.aeExposureCompensation, in counts of android.control.aeCompensationStep,
that are supported by this camera device.
Range [0,0] indicates that exposure compensation is not supported.
For LIMITED and FULL devices, range must follow below requirements if exposure
compensation is supported (`range != [0, 0]`):
`Min.exposure compensation * android.control.aeCompensationStep <= -2 EV`
`Max.exposure compensation * android.control.aeCompensationStep >= 2 EV`
LEGACY devices may support a smaller range than this.
Smallest step by which the exposure compensation
can be changed.
Exposure Value (EV)
This is the unit for android.control.aeExposureCompensation. For example, if this key has
a value of `1/2`, then a setting of `-2` for android.control.aeExposureCompensation means
that the target EV offset for the auto-exposure routine is -1 EV.
One unit of EV compensation changes the brightness of the captured image by a factor
of two. +1 EV doubles the image brightness, while -1 EV halves the image brightness.
This must be less than or equal to 1/2.
n
List of auto-focus (AF) modes for android.control.afMode that are
supported by this camera device.
Any value listed in android.control.afMode
Not all the auto-focus modes may be supported by a
given camera device. This entry lists the valid modes for
android.control.afMode for this camera device.
All LIMITED and FULL mode camera devices will support OFF mode, and all
camera devices with adjustable focuser units
(`android.lens.info.minimumFocusDistance > 0`) will support AUTO mode.
LEGACY devices will support OFF mode only if they support
focusing to infinity (by also setting android.lens.focusDistance to
`0.0f`).
n
List of color effects for android.control.effectMode that are supported by this camera
device.
Any value listed in android.control.effectMode
This list contains the color effect modes that can be applied to
images produced by the camera device.
Implementations are not expected to be consistent across all devices.
If no color effect modes are available for a device, this will only list
OFF.
A color effect will only be applied if
android.control.mode != OFF. OFF is always included in this list.
This control has no effect on the operation of other control routines such
as auto-exposure, white balance, or focus.
n
List of scene modes for android.control.sceneMode that are supported by this camera
device.
Any value listed in android.control.sceneMode
This list contains scene modes that can be set for the camera device.
Only scene modes that have been fully implemented for the
camera device may be included here. Implementations are not expected
to be consistent across all devices.
If no scene modes are supported by the camera device, this
will be set to DISABLED. Otherwise DISABLED will not be listed.
FACE_PRIORITY is always listed if face detection is
supported (i.e.`android.statistics.info.maxFaceCount >
0`).
n
List of video stabilization modes for android.control.videoStabilizationMode
that are supported by this camera device.
Any value listed in android.control.videoStabilizationMode
OFF will always be listed.
n
List of auto-white-balance modes for android.control.awbMode that are supported by this
camera device.
Any value listed in android.control.awbMode
Not all the auto-white-balance modes may be supported by a
given camera device. This entry lists the valid modes for
android.control.awbMode for this camera device.
All camera devices will support ON mode.
Camera devices that support the MANUAL_POST_PROCESSING capability will always support OFF
mode, which enables application control of white balance, by using
android.colorCorrection.transform and android.colorCorrection.gains
(android.colorCorrection.mode must be set to TRANSFORM_MATRIX). This includes all FULL
mode camera devices.
3
List of the maximum number of regions that can be used for metering in
auto-exposure (AE), auto-white balance (AWB), and auto-focus (AF);
this corresponds to the the maximum number of elements in
android.control.aeRegions, android.control.awbRegions,
and android.control.afRegions.
Value must be >= 0 for each element. For full-capability devices
this value must be >= 1 for AE and AF. The order of the elements is:
`(AE, AWB, AF)`.
The maximum number of metering regions that can be used by the auto-exposure (AE)
routine.
Value will be >= 0. For FULL-capability devices, this
value will be >= 1.
This corresponds to the the maximum allowed number of elements in
android.control.aeRegions.
This entry is private to the framework. Fill in
maxRegions to have this entry be automatically populated.
The maximum number of metering regions that can be used by the auto-white balance (AWB)
routine.
Value will be >= 0.
This corresponds to the the maximum allowed number of elements in
android.control.awbRegions.
This entry is private to the framework. Fill in
maxRegions to have this entry be automatically populated.
The maximum number of metering regions that can be used by the auto-focus (AF) routine.
Value will be >= 0. For FULL-capability devices, this
value will be >= 1.
This corresponds to the the maximum allowed number of elements in
android.control.afRegions.
This entry is private to the framework. Fill in
maxRegions to have this entry be automatically populated.
3
length(availableSceneModes)
Ordered list of auto-exposure, auto-white balance, and auto-focus
settings to use with each available scene mode.
For each available scene mode, the list must contain three
entries containing the android.control.aeMode,
android.control.awbMode, and android.control.afMode values used
by the camera device. The entry order is `(aeMode, awbMode, afMode)`
where aeMode has the lowest index position.
When a scene mode is enabled, the camera device is expected
to override android.control.aeMode, android.control.awbMode,
and android.control.afMode with its preferred settings for
that scene mode.
The order of this list matches that of availableSceneModes,
with 3 entries for each mode. The overrides listed
for FACE_PRIORITY and FACE_PRIORITY_LOW_LIGHT (if supported) are ignored,
since for that mode the application-set android.control.aeMode,
android.control.awbMode, and android.control.afMode values are
used instead, matching the behavior when android.control.mode
is set to AUTO. It is recommended that the FACE_PRIORITY and
FACE_PRIORITY_LOW_LIGHT (if supported) overrides should be set to 0.
For example, if availableSceneModes contains
`(FACE_PRIORITY, ACTION, NIGHT)`, then the camera framework
expects sceneModeOverrides to have 9 entries formatted like:
`(0, 0, 0, ON_AUTO_FLASH, AUTO, CONTINUOUS_PICTURE,
ON_AUTO_FLASH, INCANDESCENT, AUTO)`.
To maintain backward compatibility, this list will be made available
in the static metadata of the camera service. The camera service will
use these values to set android.control.aeMode,
android.control.awbMode, and android.control.afMode when using a scene
mode other than FACE_PRIORITY and FACE_PRIORITY_LOW_LIGHT (if supported).
The ID sent with the latest
CAMERA2_TRIGGER_PRECAPTURE_METERING call
Removed in camera HAL v3
Must be 0 if no
CAMERA2_TRIGGER_PRECAPTURE_METERING trigger received yet
by HAL. Always updated even if AE algorithm ignores the
trigger
INACTIVE
AE is off or recently reset.
When a camera device is opened, it starts in
this state. This is a transient state, the camera device may skip reporting
this state in capture result.
SEARCHING
AE doesn't yet have a good set of control values
for the current scene.
This is a transient state, the camera device may skip
reporting this state in capture result.
CONVERGED
AE has a good set of control values for the
current scene.
LOCKED
AE has been locked.
FLASH_REQUIRED
AE has a good set of control values, but flash
needs to be fired for good quality still
capture.
PRECAPTURE
AE has been asked to do a precapture sequence
and is currently executing it.
Precapture can be triggered through setting
android.control.aePrecaptureTrigger to START. Currently
active and completed (if it causes camera device internal AE lock) precapture
metering sequence can be canceled through setting
android.control.aePrecaptureTrigger to CANCEL.
Once PRECAPTURE completes, AE will transition to CONVERGED
or FLASH_REQUIRED as appropriate. This is a transient
state, the camera device may skip reporting this state in
capture result.
Current state of the auto-exposure (AE) algorithm.
Switching between or enabling AE modes (android.control.aeMode) always
resets the AE state to INACTIVE. Similarly, switching between android.control.mode,
or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all
the algorithm states to INACTIVE.
The camera device can do several state transitions between two results, if it is
allowed by the state transition table. For example: INACTIVE may never actually be
seen in a result.
The state in the result is the state for this image (in sync with this image): if
AE state becomes CONVERGED, then the image data associated with this result should
be good to use.
Below are state transition tables for different AE modes.
State | Transition Cause | New State | Notes
:------------:|:----------------:|:---------:|:-----------------------:
INACTIVE | | INACTIVE | Camera device auto exposure algorithm is disabled
When android.control.aeMode is AE_MODE_ON*:
State | Transition Cause | New State | Notes
:-------------:|:--------------------------------------------:|:--------------:|:-----------------:
INACTIVE | Camera device initiates AE scan | SEARCHING | Values changing
INACTIVE | android.control.aeLock is ON | LOCKED | Values locked
SEARCHING | Camera device finishes AE scan | CONVERGED | Good values, not changing
SEARCHING | Camera device finishes AE scan | FLASH_REQUIRED | Converged but too dark w/o flash
SEARCHING | android.control.aeLock is ON | LOCKED | Values locked
CONVERGED | Camera device initiates AE scan | SEARCHING | Values changing
CONVERGED | android.control.aeLock is ON | LOCKED | Values locked
FLASH_REQUIRED | Camera device initiates AE scan | SEARCHING | Values changing
FLASH_REQUIRED | android.control.aeLock is ON | LOCKED | Values locked
LOCKED | android.control.aeLock is OFF | SEARCHING | Values not good after unlock
LOCKED | android.control.aeLock is OFF | CONVERGED | Values good after unlock
LOCKED | android.control.aeLock is OFF | FLASH_REQUIRED | Exposure good, but too dark
PRECAPTURE | Sequence done. android.control.aeLock is OFF | CONVERGED | Ready for high-quality capture
PRECAPTURE | Sequence done. android.control.aeLock is ON | LOCKED | Ready for high-quality capture
LOCKED | aeLock is ON and aePrecaptureTrigger is START | LOCKED | Precapture trigger is ignored when AE is already locked
LOCKED | aeLock is ON and aePrecaptureTrigger is CANCEL| LOCKED | Precapture trigger is ignored when AE is already locked
Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is START | PRECAPTURE | Start AE precapture metering sequence
Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is CANCEL| INACTIVE | Currently active precapture metering sequence is canceled
If the camera device supports AE external flash mode (ON_EXTERNAL_FLASH is included in
android.control.aeAvailableModes), android.control.aeState must be FLASH_REQUIRED after
the camera device finishes AE scan and it's too dark without flash.
For the above table, the camera device may skip reporting any state changes that happen
without application intervention (i.e. mode switch, trigger, locking). Any state that
can be skipped in that manner is called a transient state.
For example, for above AE modes (AE_MODE_ON*), in addition to the state transitions
listed in above table, it is also legal for the camera device to skip one or more
transient states between two results. See below table for examples:
State | Transition Cause | New State | Notes
:-------------:|:-----------------------------------------------------------:|:--------------:|:-----------------:
INACTIVE | Camera device finished AE scan | CONVERGED | Values are already good, transient states are skipped by camera device.
Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is START, sequence done | FLASH_REQUIRED | Converged but too dark w/o flash after a precapture sequence, transient states are skipped by camera device.
Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is START, sequence done | CONVERGED | Converged after a precapture sequence, transient states are skipped by camera device.
Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is CANCEL, converged | FLASH_REQUIRED | Converged but too dark w/o flash after a precapture sequence is canceled, transient states are skipped by camera device.
Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is CANCEL, converged | CONVERGED | Converged after a precapture sequenceis canceled, transient states are skipped by camera device.
CONVERGED | Camera device finished AE scan | FLASH_REQUIRED | Converged but too dark w/o flash after a new scan, transient states are skipped by camera device.
FLASH_REQUIRED | Camera device finished AE scan | CONVERGED | Converged after a new scan, transient states are skipped by camera device.
INACTIVE
AF is off or has not yet tried to scan/been asked
to scan.
When a camera device is opened, it starts in this
state. This is a transient state, the camera device may
skip reporting this state in capture
result.
PASSIVE_SCAN
AF is currently performing an AF scan initiated the
camera device in a continuous autofocus mode.
Only used by CONTINUOUS_* AF modes. This is a transient
state, the camera device may skip reporting this state in
capture result.
PASSIVE_FOCUSED
AF currently believes it is in focus, but may
restart scanning at any time.
Only used by CONTINUOUS_* AF modes. This is a transient
state, the camera device may skip reporting this state in
capture result.
ACTIVE_SCAN
AF is performing an AF scan because it was
triggered by AF trigger.
Only used by AUTO or MACRO AF modes. This is a transient
state, the camera device may skip reporting this state in
capture result.
FOCUSED_LOCKED
AF believes it is focused correctly and has locked
focus.
This state is reached only after an explicit START AF trigger has been
sent (android.control.afTrigger), when good focus has been obtained.
The lens will remain stationary until the AF mode (android.control.afMode) is changed or
a new AF trigger is sent to the camera device (android.control.afTrigger).
NOT_FOCUSED_LOCKED
AF has failed to focus successfully and has locked
focus.
This state is reached only after an explicit START AF trigger has been
sent (android.control.afTrigger), when good focus cannot be obtained.
The lens will remain stationary until the AF mode (android.control.afMode) is changed or
a new AF trigger is sent to the camera device (android.control.afTrigger).
PASSIVE_UNFOCUSED
AF finished a passive scan without finding focus,
and may restart scanning at any time.
Only used by CONTINUOUS_* AF modes. This is a transient state, the camera
device may skip reporting this state in capture result.
LEGACY camera devices do not support this state. When a passive
scan has finished, it will always go to PASSIVE_FOCUSED.
Current state of auto-focus (AF) algorithm.
Switching between or enabling AF modes (android.control.afMode) always
resets the AF state to INACTIVE. Similarly, switching between android.control.mode,
or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all
the algorithm states to INACTIVE.
The camera device can do several state transitions between two results, if it is
allowed by the state transition table. For example: INACTIVE may never actually be
seen in a result.
The state in the result is the state for this image (in sync with this image): if
AF state becomes FOCUSED, then the image data associated with this result should
be sharp.
Below are state transition tables for different AF modes.
When android.control.afMode is AF_MODE_OFF or AF_MODE_EDOF:
State | Transition Cause | New State | Notes
:------------:|:----------------:|:---------:|:-----------:
INACTIVE | | INACTIVE | Never changes
When android.control.afMode is AF_MODE_AUTO or AF_MODE_MACRO:
State | Transition Cause | New State | Notes
:-----------------:|:----------------:|:------------------:|:--------------:
INACTIVE | AF_TRIGGER | ACTIVE_SCAN | Start AF sweep, Lens now moving
ACTIVE_SCAN | AF sweep done | FOCUSED_LOCKED | Focused, Lens now locked
ACTIVE_SCAN | AF sweep done | NOT_FOCUSED_LOCKED | Not focused, Lens now locked
ACTIVE_SCAN | AF_CANCEL | INACTIVE | Cancel/reset AF, Lens now locked
FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Cancel/reset AF
FOCUSED_LOCKED | AF_TRIGGER | ACTIVE_SCAN | Start new sweep, Lens now moving
NOT_FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Cancel/reset AF
NOT_FOCUSED_LOCKED | AF_TRIGGER | ACTIVE_SCAN | Start new sweep, Lens now moving
Any state | Mode change | INACTIVE |
For the above table, the camera device may skip reporting any state changes that happen
without application intervention (i.e. mode switch, trigger, locking). Any state that
can be skipped in that manner is called a transient state.
For example, for these AF modes (AF_MODE_AUTO and AF_MODE_MACRO), in addition to the
state transitions listed in above table, it is also legal for the camera device to skip
one or more transient states between two results. See below table for examples:
State | Transition Cause | New State | Notes
:-----------------:|:----------------:|:------------------:|:--------------:
INACTIVE | AF_TRIGGER | FOCUSED_LOCKED | Focus is already good or good after a scan, lens is now locked.
INACTIVE | AF_TRIGGER | NOT_FOCUSED_LOCKED | Focus failed after a scan, lens is now locked.
FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | Focus is already good or good after a scan, lens is now locked.
NOT_FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | Focus is good after a scan, lens is not locked.
When android.control.afMode is AF_MODE_CONTINUOUS_VIDEO:
State | Transition Cause | New State | Notes
:-----------------:|:-----------------------------------:|:------------------:|:--------------:
INACTIVE | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
INACTIVE | AF_TRIGGER | NOT_FOCUSED_LOCKED | AF state query, Lens now locked
PASSIVE_SCAN | Camera device completes current scan| PASSIVE_FOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | Camera device fails current scan | PASSIVE_UNFOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | AF_TRIGGER | FOCUSED_LOCKED | Immediate transition, if focus is good. Lens now locked
PASSIVE_SCAN | AF_TRIGGER | NOT_FOCUSED_LOCKED | Immediate transition, if focus is bad. Lens now locked
PASSIVE_SCAN | AF_CANCEL | INACTIVE | Reset lens position, Lens now locked
PASSIVE_FOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_UNFOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_FOCUSED | AF_TRIGGER | FOCUSED_LOCKED | Immediate transition, lens now locked
PASSIVE_UNFOCUSED | AF_TRIGGER | NOT_FOCUSED_LOCKED | Immediate transition, lens now locked
FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | No effect
FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
NOT_FOCUSED_LOCKED | AF_TRIGGER | NOT_FOCUSED_LOCKED | No effect
NOT_FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
When android.control.afMode is AF_MODE_CONTINUOUS_PICTURE:
State | Transition Cause | New State | Notes
:-----------------:|:------------------------------------:|:------------------:|:--------------:
INACTIVE | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
INACTIVE | AF_TRIGGER | NOT_FOCUSED_LOCKED | AF state query, Lens now locked
PASSIVE_SCAN | Camera device completes current scan | PASSIVE_FOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | Camera device fails current scan | PASSIVE_UNFOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | AF_TRIGGER | FOCUSED_LOCKED | Eventual transition once the focus is good. Lens now locked
PASSIVE_SCAN | AF_TRIGGER | NOT_FOCUSED_LOCKED | Eventual transition if cannot find focus. Lens now locked
PASSIVE_SCAN | AF_CANCEL | INACTIVE | Reset lens position, Lens now locked
PASSIVE_FOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_UNFOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_FOCUSED | AF_TRIGGER | FOCUSED_LOCKED | Immediate trans. Lens now locked
PASSIVE_UNFOCUSED | AF_TRIGGER | NOT_FOCUSED_LOCKED | Immediate trans. Lens now locked
FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | No effect
FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
NOT_FOCUSED_LOCKED | AF_TRIGGER | NOT_FOCUSED_LOCKED | No effect
NOT_FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
When switch between AF_MODE_CONTINUOUS_* (CAF modes) and AF_MODE_AUTO/AF_MODE_MACRO
(AUTO modes), the initial INACTIVE or PASSIVE_SCAN states may be skipped by the
camera device. When a trigger is included in a mode switch request, the trigger
will be evaluated in the context of the new mode in the request.
See below table for examples:
State | Transition Cause | New State | Notes
:-----------:|:--------------------------------------:|:----------------------------------------:|:--------------:
any state | CAF-->AUTO mode switch | INACTIVE | Mode switch without trigger, initial state must be INACTIVE
any state | CAF-->AUTO mode switch with AF_TRIGGER | trigger-reachable states from INACTIVE | Mode switch with trigger, INACTIVE is skipped
any state | AUTO-->CAF mode switch | passively reachable states from INACTIVE | Mode switch without trigger, passive transient state is skipped
The ID sent with the latest
CAMERA2_TRIGGER_AUTOFOCUS call
Removed in camera HAL v3
Must be 0 if no CAMERA2_TRIGGER_AUTOFOCUS trigger
received yet by HAL. Always updated even if AF algorithm
ignores the trigger
INACTIVE
AWB is not in auto mode, or has not yet started metering.
When a camera device is opened, it starts in this
state. This is a transient state, the camera device may
skip reporting this state in capture
result.
SEARCHING
AWB doesn't yet have a good set of control
values for the current scene.
This is a transient state, the camera device
may skip reporting this state in capture result.
CONVERGED
AWB has a good set of control values for the
current scene.
LOCKED
AWB has been locked.
Current state of auto-white balance (AWB) algorithm.
Switching between or enabling AWB modes (android.control.awbMode) always
resets the AWB state to INACTIVE. Similarly, switching between android.control.mode,
or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all
the algorithm states to INACTIVE.
The camera device can do several state transitions between two results, if it is
allowed by the state transition table. So INACTIVE may never actually be seen in
a result.
The state in the result is the state for this image (in sync with this image): if
AWB state becomes CONVERGED, then the image data associated with this result should
be good to use.
Below are state transition tables for different AWB modes.
When `android.control.awbMode != AWB_MODE_AUTO`:
State | Transition Cause | New State | Notes
:------------:|:----------------:|:---------:|:-----------------------:
INACTIVE | |INACTIVE |Camera device auto white balance algorithm is disabled
When android.control.awbMode is AWB_MODE_AUTO:
State | Transition Cause | New State | Notes
:-------------:|:--------------------------------:|:-------------:|:-----------------:
INACTIVE | Camera device initiates AWB scan | SEARCHING | Values changing
INACTIVE | android.control.awbLock is ON | LOCKED | Values locked
SEARCHING | Camera device finishes AWB scan | CONVERGED | Good values, not changing
SEARCHING | android.control.awbLock is ON | LOCKED | Values locked
CONVERGED | Camera device initiates AWB scan | SEARCHING | Values changing
CONVERGED | android.control.awbLock is ON | LOCKED | Values locked
LOCKED | android.control.awbLock is OFF | SEARCHING | Values not good after unlock
For the above table, the camera device may skip reporting any state changes that happen
without application intervention (i.e. mode switch, trigger, locking). Any state that
can be skipped in that manner is called a transient state.
For example, for this AWB mode (AWB_MODE_AUTO), in addition to the state transitions
listed in above table, it is also legal for the camera device to skip one or more
transient states between two results. See below table for examples:
State | Transition Cause | New State | Notes
:-------------:|:--------------------------------:|:-------------:|:-----------------:
INACTIVE | Camera device finished AWB scan | CONVERGED | Values are already good, transient states are skipped by camera device.
LOCKED | android.control.awbLock is OFF | CONVERGED | Values good after unlock, transient states are skipped by camera device.
5
n
List of available high speed video size, fps range and max batch size configurations
supported by the camera device, in the format of (width, height, fps_min, fps_max, batch_size_max).
For each configuration, the fps_max >= 120fps.
When CONSTRAINED_HIGH_SPEED_VIDEO is supported in android.request.availableCapabilities,
this metadata will list the supported high speed video size, fps range and max batch size
configurations. All the sizes listed in this configuration will be a subset of the sizes
reported by {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes}
for processed non-stalling formats.
For the high speed video use case, the application must
select the video size and fps range from this metadata to configure the recording and
preview streams and setup the recording requests. For example, if the application intends
to do high speed recording, it can select the maximum size reported by this metadata to
configure output streams. Once the size is selected, application can filter this metadata
by selected size and get the supported fps ranges, and use these fps ranges to setup the
recording requests. Note that for the use case of multiple output streams, application
must select one unique size from this metadata to use (e.g., preview and recording streams
must have the same size). Otherwise, the high speed capture session creation will fail.
The min and max fps will be multiple times of 30fps.
High speed video streaming extends significant performance pressue to camera hardware,
to achieve efficient high speed streaming, the camera device may have to aggregate
multiple frames together and send to camera device for processing where the request
controls are same for all the frames in this batch. Max batch size indicates
the max possible number of frames the camera device will group together for this high
speed stream configuration. This max batch size will be used to generate a high speed
recording request list by
{@link android.hardware.camera2.CameraConstrainedHighSpeedCaptureSession#createHighSpeedRequestList}.
The max batch size for each configuration will satisfy below conditions:
* Each max batch size will be a divisor of its corresponding fps_max / 30. For example,
if max_fps is 300, max batch size will only be 1, 2, 5, or 10.
* The camera device may choose smaller internal batch size for each configuration, but
the actual batch size will be a divisor of max batch size. For example, if the max batch
size is 8, the actual batch size used by camera device will only be 1, 2, 4, or 8.
* The max batch size in each configuration entry must be no larger than 32.
The camera device doesn't have to support batch mode to achieve high speed video recording,
in such case, batch_size_max will be reported as 1 in each configuration entry.
This fps ranges in this configuration list can only be used to create requests
that are submitted to a high speed camera capture session created by
{@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}.
The fps ranges reported in this metadata must not be used to setup capture requests for
normal capture session, or it will cause request error.
All the sizes listed in this configuration will be a subset of the sizes reported by
android.scaler.availableStreamConfigurations for processed non-stalling output formats.
Note that for all high speed video configurations, HAL must be able to support a minimum
of two streams, though the application might choose to configure just one stream.
The HAL may support multiple sensor modes for high speed outputs, for example, 120fps
sensor mode and 120fps recording, 240fps sensor mode for 240fps recording. The application
usually starts preview first, then starts recording. To avoid sensor mode switch caused
stutter when starting recording as much as possible, the application may want to ensure
the same sensor mode is used for preview and recording. Therefore, The HAL must advertise
the variable fps range [30, fps_max] for each fixed fps range in this configuration list.
For example, if the HAL advertises [120, 120] and [240, 240], the HAL must also advertise
[30, 120] and [30, 240] for each configuration. In doing so, if the application intends to
do 120fps recording, it can select [30, 120] to start preview, and [120, 120] to start
recording. For these variable fps ranges, it's up to the HAL to decide the actual fps
values that are suitable for smooth preview streaming. If the HAL sees different max_fps
values that fall into different sensor modes in a sequence of requests, the HAL must
switch the sensor mode as quick as possible to minimize the mode switch caused stutter.
FALSE
TRUE
Whether the camera device supports android.control.aeLock
Devices with MANUAL_SENSOR capability or BURST_CAPTURE capability will always
list `true`. This includes FULL devices.
FALSE
TRUE
Whether the camera device supports android.control.awbLock
Devices with MANUAL_POST_PROCESSING capability or BURST_CAPTURE capability will
always list `true`. This includes FULL devices.
n
List of control modes for android.control.mode that are supported by this camera
device.
Any value listed in android.control.mode
This list contains control modes that can be set for the camera device.
LEGACY mode devices will always support AUTO mode. LIMITED and FULL
devices will always support OFF, AUTO modes.
2
Range of boosts for android.control.postRawSensitivityBoost supported
by this camera device.
ISO arithmetic units, the same as android.sensor.sensitivity
Devices support post RAW sensitivity boost will advertise
android.control.postRawSensitivityBoost key for controling
post RAW sensitivity boost.
This key will be `null` for devices that do not support any RAW format
outputs. For devices that do support RAW format outputs, this key will always
present, and if a device does not support post RAW sensitivity boost, it will
list `(100, 100)` in this key.
This key is added in legacy HAL3.4. For legacy HAL3.3 or earlier devices, camera
framework will generate this key as `(100, 100)` if device supports any of RAW output
formats. All legacy HAL3.4 and above devices should list this key if device supports
any of RAW output formats.
The amount of additional sensitivity boost applied to output images
after RAW sensor data is captured.
ISO arithmetic units, the same as android.sensor.sensitivity
android.control.postRawSensitivityBoostRange
Some camera devices support additional digital sensitivity boosting in the
camera processing pipeline after sensor RAW image is captured.
Such a boost will be applied to YUV/JPEG format output images but will not
have effect on RAW output formats like RAW_SENSOR, RAW10, RAW12 or RAW_OPAQUE.
This key will be `null` for devices that do not support any RAW format
outputs. For devices that do support RAW format outputs, this key will always
present, and if a device does not support post RAW sensitivity boost, it will
list `100` in this key.
If the camera device cannot apply the exact boost requested, it will reduce the
boost to the nearest supported value.
The final boost value used will be available in the output capture result.
For devices that support post RAW sensitivity boost, the YUV/JPEG output images
of such device will have the total sensitivity of
`android.sensor.sensitivity * android.control.postRawSensitivityBoost / 100`
The sensitivity of RAW format images will always be `android.sensor.sensitivity`
This control is only effective if android.control.aeMode or android.control.mode is set to
OFF; otherwise the auto-exposure algorithm will override this value.
FALSE
Requests with android.control.captureIntent == STILL_CAPTURE must be captured
after previous requests.
TRUE
Requests with android.control.captureIntent == STILL_CAPTURE may or may not be
captured before previous requests.
Allow camera device to enable zero-shutter-lag mode for requests with
android.control.captureIntent == STILL_CAPTURE.
If enableZsl is `true`, the camera device may enable zero-shutter-lag mode for requests with
STILL_CAPTURE capture intent. The camera device may use images captured in the past to
produce output images for a zero-shutter-lag request. The result metadata including the
android.sensor.timestamp reflects the source frames used to produce output images.
Therefore, the contents of the output images and the result metadata may be out of order
compared to previous regular requests. enableZsl does not affect requests with other
capture intents.
For example, when requests are submitted in the following order:
Request A: enableZsl is ON, android.control.captureIntent is PREVIEW
Request B: enableZsl is ON, android.control.captureIntent is STILL_CAPTURE
The output images for request B may have contents captured before the output images for
request A, and the result metadata for request B may be older than the result metadata for
request A.
Note that when enableZsl is `true`, it is not guaranteed to get output images captured in
the past for requests with STILL_CAPTURE capture intent.
For applications targeting SDK versions O and newer, the value of enableZsl in
TEMPLATE_STILL_CAPTURE template may be `true`. The value in other templates is always
`false` if present.
For applications targeting SDK versions older than O, the value of enableZsl in all
capture templates is always `false` if present.
For application-operated ZSL, use CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.
It is valid for HAL to produce regular output images for requests with STILL_CAPTURE
capture intent.
NOT_DETECTED
Scene change is not detected within the AF region(s).
DETECTED
Scene change is detected within the AF region(s).
Whether a significant scene change is detected within the currently-set AF
region(s).
When the camera focus routine detects a change in the scene it is looking at,
such as a large shift in camera viewpoint, significant motion in the scene, or a
significant illumination change, this value will be set to DETECTED for a single capture
result. Otherwise the value will be NOT_DETECTED. The threshold for detection is similar
to what would trigger a new passive focus scan to begin in CONTINUOUS autofocus modes.
This key will be available if the camera device advertises this key via {@link
android.hardware.camera2.CameraCharacteristics#getAvailableCaptureResultKeys|ACAMERA_REQUEST_AVAILABLE_RESULT_KEYS}.
FAST
Minimal or no slowdown of frame rate compared to
Bayer RAW output.
HIGH_QUALITY
Improved processing quality but the frame rate might be slowed down
relative to raw output.
Controls the quality of the demosaicing
processing.
OFF
No edge enhancement is applied.
FAST
Apply edge enhancement at a quality level that does not slow down frame rate
relative to sensor output. It may be the same as OFF if edge enhancement will
slow down frame rate relative to sensor.
HIGH_QUALITY
Apply high-quality edge enhancement, at a cost of possibly reduced output frame rate.
ZERO_SHUTTER_LAG Edge enhancement is applied at different
levels for different output streams, based on resolution. Streams at maximum recording
resolution (see {@link
android.hardware.camera2.CameraDevice#createCaptureSession|ACameraDevice_createCaptureSession})
or below have edge enhancement applied, while higher-resolution streams have no edge
enhancement applied. The level of edge enhancement for low-resolution streams is tuned
so that frame rate is not impacted, and the quality is equal to or better than FAST
(since it is only applied to lower-resolution outputs, quality may improve from FAST).
This mode is intended to be used by applications operating in a zero-shutter-lag mode
with YUV or PRIVATE reprocessing, where the application continuously captures
high-resolution intermediate buffers into a circular buffer, from which a final image is
produced via reprocessing when a user takes a picture. For such a use case, the
high-resolution buffers must not have edge enhancement applied to maximize efficiency of
preview and to avoid double-applying enhancement when reprocessed, while low-resolution
buffers (used for recording or preview, generally) need edge enhancement applied for
reasonable preview quality.
This mode is guaranteed to be supported by devices that support either the
YUV_REPROCESSING or PRIVATE_REPROCESSING capabilities
(android.request.availableCapabilities lists either of those capabilities) and it will
be the default mode for CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.
Operation mode for edge
enhancement.
android.edge.availableEdgeModes
Edge enhancement improves sharpness and details in the captured image. OFF means
no enhancement will be applied by the camera device.
FAST/HIGH_QUALITY both mean camera device determined enhancement
will be applied. HIGH_QUALITY mode indicates that the
camera device will use the highest-quality enhancement algorithms,
even if it slows down capture rate. FAST means the camera device will
not slow down capture rate when applying edge enhancement. FAST may be the same as OFF if
edge enhancement will slow down capture rate. Every output stream will have a similar
amount of enhancement applied.
ZERO_SHUTTER_LAG is meant to be used by applications that maintain a continuous circular
buffer of high-resolution images during preview and reprocess image(s) from that buffer
into a final capture when triggered by the user. In this mode, the camera device applies
edge enhancement to low-resolution streams (below maximum recording resolution) to
maximize preview quality, but does not apply edge enhancement to high-resolution streams,
since those will be reprocessed later if necessary.
For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera
device will apply FAST/HIGH_QUALITY YUV-domain edge enhancement, respectively.
The camera device may adjust its internal edge enhancement parameters for best
image quality based on the android.reprocess.effectiveExposureFactor, if it is set.
For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to
adjust the internal edge enhancement reduction parameters appropriately to get the best
quality images.
Control the amount of edge enhancement
applied to the images
1-10; 10 is maximum sharpening
n
List of edge enhancement modes for android.edge.mode that are supported by this camera
device.
Any value listed in android.edge.mode
Full-capability camera devices must always support OFF; camera devices that support
YUV_REPROCESSING or PRIVATE_REPROCESSING will list ZERO_SHUTTER_LAG; all devices will
list FAST.
HAL must support both FAST and HIGH_QUALITY if edge enhancement control is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY will generate the same output.
Power for flash firing/torch
10 is max power; 0 is no flash. Linear
0 - 10
Power for snapshot may use a different scale than
for torch mode. Only one entry for torch mode will be
used
Firing time of flash relative to start of
exposure
nanoseconds
0-(exposure time-flash duration)
Clamped to (0, exposure time - flash
duration).
OFF
Do not fire the flash for this capture.
SINGLE
If the flash is available and charged, fire flash
for this capture.
TORCH
Transition flash to continuously on.
The desired mode for for the camera device's flash control.
This control is only effective when flash unit is available
(`android.flash.info.available == true`).
When this control is used, the android.control.aeMode must be set to ON or OFF.
Otherwise, the camera device auto-exposure related flash control (ON_AUTO_FLASH,
ON_ALWAYS_FLASH, or ON_AUTO_FLASH_REDEYE) will override this control.
When set to OFF, the camera device will not fire flash for this capture.
When set to SINGLE, the camera device will fire flash regardless of the camera
device's auto-exposure routine's result. When used in still capture case, this
control should be used along with auto-exposure (AE) precapture metering sequence
(android.control.aePrecaptureTrigger), otherwise, the image may be incorrectly exposed.
When set to TORCH, the flash will be on continuously. This mode can be used
for use cases such as preview, auto-focus assist, still capture, or video recording.
The flash status will be reported by android.flash.state in the capture result metadata.
FALSE
TRUE
Whether this camera device has a
flash unit.
Will be `false` if no flash is available.
If there is no flash unit, none of the flash controls do
anything.
Time taken before flash can fire
again
nanoseconds
0-1e9
1 second too long/too short for recharge? Should
this be power-dependent?
The x,y whitepoint of the
flash
pair of floats
0-1 for both
Max energy output of the flash for a full
power single flash
lumen-seconds
>= 0
UNAVAILABLE
No flash on camera.
CHARGING
Flash is charging and cannot be fired.
READY
Flash is ready to fire.
FIRED
Flash fired for this capture.
PARTIAL
Flash partially illuminated this frame.
This is usually due to the next or previous frame having
the flash fire, and the flash spilling into this capture
due to hardware limitations.
Current state of the flash
unit.
When the camera device doesn't have flash unit
(i.e. `android.flash.info.available == false`), this state will always be UNAVAILABLE.
Other states indicate the current flash status.
In certain conditions, this will be available on LEGACY devices:
* Flash-less cameras always return UNAVAILABLE.
* Using android.control.aeMode `==` ON_ALWAYS_FLASH
will always return FIRED.
* Using android.flash.mode `==` TORCH
will always return FIRED.
In all other conditions the state will not be available on
LEGACY devices (i.e. it will be `null`).
OFF
No hot pixel correction is applied.
The frame rate must not be reduced relative to sensor raw output
for this option.
The hotpixel map may be returned in android.statistics.hotPixelMap.
FAST
Hot pixel correction is applied, without reducing frame
rate relative to sensor raw output.
The hotpixel map may be returned in android.statistics.hotPixelMap.
HIGH_QUALITY
High-quality hot pixel correction is applied, at a cost
of possibly reduced frame rate relative to sensor raw output.
The hotpixel map may be returned in android.statistics.hotPixelMap.
Operational mode for hot pixel correction.
android.hotPixel.availableHotPixelModes
Hotpixel correction interpolates out, or otherwise removes, pixels
that do not accurately measure the incoming light (i.e. pixels that
are stuck at an arbitrary value or are oversensitive).
n
List of hot pixel correction modes for android.hotPixel.mode that are supported by this
camera device.
Any value listed in android.hotPixel.mode
FULL mode camera devices will always support FAST.
To avoid performance issues, there will be significantly fewer hot
pixels than actual pixels on the camera sensor.
HAL must support both FAST and HIGH_QUALITY if hot pixel correction control is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY will generate the same output.
A location object to use when generating image GPS metadata.
Setting a location object in a request will include the GPS coordinates of the location
into any JPEG images captured based on the request. These coordinates can then be
viewed by anyone who receives the JPEG image.
This tag is also used for HEIC image capture.
3
GPS coordinates to include in output JPEG
EXIF.
(-180 - 180], [-90,90], [-inf, inf]
This tag is also used for HEIC image capture.
32 characters describing GPS algorithm to
include in EXIF.
UTF-8 null-terminated string
This tag is also used for HEIC image capture.
Time GPS fix was made to include in
EXIF.
UTC in seconds since January 1, 1970
This tag is also used for HEIC image capture.
The orientation for a JPEG image.
Degrees in multiples of 90
0, 90, 180, 270
The clockwise rotation angle in degrees, relative to the orientation
to the camera, that the JPEG picture needs to be rotated by, to be viewed
upright.
Camera devices may either encode this value into the JPEG EXIF header, or
rotate the image data to match this orientation. When the image data is rotated,
the thumbnail data will also be rotated.
Note that this orientation is relative to the orientation of the camera sensor, given
by android.sensor.orientation.
To translate from the device orientation given by the Android sensor APIs for camera
sensors which are not EXTERNAL, the following sample code may be used:
private int getJpegOrientation(CameraCharacteristics c, int deviceOrientation) {
if (deviceOrientation == android.view.OrientationEventListener.ORIENTATION_UNKNOWN) return 0;
int sensorOrientation = c.get(CameraCharacteristics.SENSOR_ORIENTATION);
// Round device orientation to a multiple of 90
deviceOrientation = (deviceOrientation + 45) / 90 * 90;
// Reverse device orientation for front-facing cameras
boolean facingFront = c.get(CameraCharacteristics.LENS_FACING) == CameraCharacteristics.LENS_FACING_FRONT;
if (facingFront) deviceOrientation = -deviceOrientation;
// Calculate desired JPEG orientation relative to camera orientation to make
// the image upright relative to the device orientation
int jpegOrientation = (sensorOrientation + deviceOrientation + 360) % 360;
return jpegOrientation;
}
For EXTERNAL cameras the sensor orientation will always be set to 0 and the facing will
also be set to EXTERNAL. The above code is not relevant in such case.
This tag is also used to describe the orientation of the HEIC image capture, in which
case the rotation is reflected by
{@link android.media.ExifInterface#TAG_ORIENTATION EXIF orientation flag}, and not by
rotating the image data itself.
Compression quality of the final JPEG
image.
1-100; larger is higher quality
85-95 is typical usage range. This tag is also used to describe the quality
of the HEIC image capture.
Compression quality of JPEG
thumbnail.
1-100; larger is higher quality
This tag is also used to describe the quality of the HEIC image capture.
2
Resolution of embedded JPEG thumbnail.
android.jpeg.availableThumbnailSizes
When set to (0, 0) value, the JPEG EXIF will not contain thumbnail,
but the captured JPEG will still be a valid image.
For best results, when issuing a request for a JPEG image, the thumbnail size selected
should have the same aspect ratio as the main JPEG output.
If the thumbnail image aspect ratio differs from the JPEG primary image aspect
ratio, the camera device creates the thumbnail by cropping it from the primary image.
For example, if the primary image has 4:3 aspect ratio, the thumbnail image has
16:9 aspect ratio, the primary image will be cropped vertically (letterbox) to
generate the thumbnail image. The thumbnail image will always have a smaller Field
Of View (FOV) than the primary image when aspect ratios differ.
When an android.jpeg.orientation of non-zero degree is requested,
the camera device will handle thumbnail rotation in one of the following ways:
* Set the {@link android.media.ExifInterface#TAG_ORIENTATION EXIF orientation flag}
and keep jpeg and thumbnail image data unrotated.
* Rotate the jpeg and thumbnail image data and not set
{@link android.media.ExifInterface#TAG_ORIENTATION EXIF orientation flag}. In this
case, LIMITED or FULL hardware level devices will report rotated thumnail size in
capture result, so the width and height will be interchanged if 90 or 270 degree
orientation is requested. LEGACY device will always report unrotated thumbnail
size.
The tag is also used as thumbnail size for HEIC image format capture, in which case the
the thumbnail rotation is reflected by
{@link android.media.ExifInterface#TAG_ORIENTATION EXIF orientation flag}, and not by
rotating the thumbnail data itself.
The HAL must not squeeze or stretch the downscaled primary image to generate thumbnail.
The cropping must be done on the primary jpeg image rather than the sensor pre-correction
active array. The stream cropping rule specified by "S5. Cropping" in camera3.h doesn't
apply to the thumbnail image cropping.
2
n
List of JPEG thumbnail sizes for android.jpeg.thumbnailSize supported by this
camera device.
This list will include at least one non-zero resolution, plus `(0,0)` for indicating no
thumbnail should be generated.
Below condiditions will be satisfied for this size list:
* The sizes will be sorted by increasing pixel area (width x height).
If several resolutions have the same area, they will be sorted by increasing width.
* The aspect ratio of the largest thumbnail size will be same as the
aspect ratio of largest JPEG output size in android.scaler.availableStreamConfigurations.
The largest size is defined as the size that has the largest pixel area
in a given size list.
* Each output JPEG size in android.scaler.availableStreamConfigurations will have at least
one corresponding size that has the same aspect ratio in availableThumbnailSizes,
and vice versa.
* All non-`(0, 0)` sizes will have non-zero widths and heights.
This list is also used as supported thumbnail sizes for HEIC image format capture.
Maximum size in bytes for the compressed
JPEG buffer
Must be large enough to fit any JPEG produced by
the camera
This is used for sizing the gralloc buffers for
JPEG
The size of the compressed JPEG image, in
bytes
>= 0
If no JPEG output is produced for the request,
this must be 0.
Otherwise, this describes the real size of the compressed
JPEG image placed in the output stream. More specifically,
if android.jpeg.maxSize = 1000000, and a specific capture
has android.jpeg.size = 500000, then the output buffer from
the JPEG stream will be 1000000 bytes, of which the first
500000 make up the real data.
The desired lens aperture size, as a ratio of lens focal length to the
effective aperture diameter.
The f-number (f/N)
android.lens.info.availableApertures
Setting this value is only supported on the camera devices that have a variable
aperture lens.
When this is supported and android.control.aeMode is OFF,
this can be set along with android.sensor.exposureTime,
android.sensor.sensitivity, and android.sensor.frameDuration
to achieve manual exposure control.
The requested aperture value may take several frames to reach the
requested value; the camera device will report the current (intermediate)
aperture size in capture result metadata while the aperture is changing.
While the aperture is still changing, android.lens.state will be set to MOVING.
When this is supported and android.control.aeMode is one of
the ON modes, this will be overridden by the camera device
auto-exposure algorithm, the overridden values are then provided
back to the user in the corresponding result.
The desired setting for the lens neutral density filter(s).
Exposure Value (EV)
android.lens.info.availableFilterDensities
This control will not be supported on most camera devices.
Lens filters are typically used to lower the amount of light the
sensor is exposed to (measured in steps of EV). As used here, an EV
step is the standard logarithmic representation, which are
non-negative, and inversely proportional to the amount of light
hitting the sensor. For example, setting this to 0 would result
in no reduction of the incoming light, and setting this to 2 would
mean that the filter is set to reduce incoming light by two stops
(allowing 1/4 of the prior amount of light to the sensor).
It may take several frames before the lens filter density changes
to the requested value. While the filter density is still changing,
android.lens.state will be set to MOVING.
The desired lens focal length; used for optical zoom.
Millimeters
android.lens.info.availableFocalLengths
This setting controls the physical focal length of the camera
device's lens. Changing the focal length changes the field of
view of the camera device, and is usually used for optical zoom.
Like android.lens.focusDistance and android.lens.aperture, this
setting won't be applied instantaneously, and it may take several
frames before the lens can change to the requested focal length.
While the focal length is still changing, android.lens.state will
be set to MOVING.
Optical zoom will not be supported on most devices.
For a logical camera device supporting both optical and digital zoom, if focalLength and
cropRegion change in the same request, the camera device must make sure that the new
focalLength and cropRegion take effect in the same frame. This is to make sure that there
is no visible field-of-view jump during zoom. For example, if cropRegion is applied
immediately, but focalLength takes more than 1 frame to take effect, the camera device
will delay the cropRegion so that it's synchronized with focalLength.
Desired distance to plane of sharpest focus,
measured from frontmost surface of the lens.
See android.lens.info.focusDistanceCalibration for details
>= 0
This control can be used for setting manual focus, on devices that support
the MANUAL_SENSOR capability and have a variable-focus lens (see
android.lens.info.minimumFocusDistance).
A value of `0.0f` means infinity focus. The value set will be clamped to
`[0.0f, android.lens.info.minimumFocusDistance]`.
Like android.lens.focalLength, this setting won't be applied
instantaneously, and it may take several frames before the lens
can move to the requested focus distance. While the lens is still moving,
android.lens.state will be set to MOVING.
LEGACY devices support at most setting this to `0.0f`
for infinity focus.
OFF
Optical stabilization is unavailable.
ON
Optical stabilization is enabled.
Sets whether the camera device uses optical image stabilization (OIS)
when capturing images.
android.lens.info.availableOpticalStabilization
OIS is used to compensate for motion blur due to small
movements of the camera during capture. Unlike digital image
stabilization (android.control.videoStabilizationMode), OIS
makes use of mechanical elements to stabilize the camera
sensor, and thus allows for longer exposure times before
camera shake becomes apparent.
Switching between different optical stabilization modes may take several
frames to initialize, the camera device will report the current mode in
capture result metadata. For example, When "ON" mode is requested, the
optical stabilization modes in the first several capture results may still
be "OFF", and it will become "ON" when the initialization is done.
If a camera device supports both OIS and digital image stabilization
(android.control.videoStabilizationMode), turning both modes on may produce undesirable
interaction, so it is recommended not to enable both at the same time.
Not all devices will support OIS; see
android.lens.info.availableOpticalStabilization for
available controls.
n
List of aperture size values for android.lens.aperture that are
supported by this camera device.
The aperture f-number
If the camera device doesn't support a variable lens aperture,
this list will contain only one value, which is the fixed aperture size.
If the camera device supports a variable aperture, the aperture values
in this list will be sorted in ascending order.
n
List of neutral density filter values for
android.lens.filterDensity that are supported by this camera device.
Exposure value (EV)
Values are >= 0
If a neutral density filter is not supported by this camera device,
this list will contain only 0. Otherwise, this list will include every
filter density supported by the camera device, in ascending order.
n
List of focal lengths for android.lens.focalLength that are supported by this camera
device.
Millimeters
Values are > 0
If optical zoom is not supported, this list will only contain
a single value corresponding to the fixed focal length of the
device. Otherwise, this list will include every focal length supported
by the camera device, in ascending order.
n
List of optical image stabilization (OIS) modes for
android.lens.opticalStabilizationMode that are supported by this camera device.
Any value listed in android.lens.opticalStabilizationMode
If OIS is not supported by a given camera device, this list will
contain only OFF.
Hyperfocal distance for this lens.
See android.lens.info.focusDistanceCalibration for details
If lens is fixed focus, >= 0. If lens has focuser unit, the value is
within `(0.0f, android.lens.info.minimumFocusDistance]`
If the lens is not fixed focus, the camera device will report this
field when android.lens.info.focusDistanceCalibration is APPROXIMATE or CALIBRATED.
Shortest distance from frontmost surface
of the lens that can be brought into sharp focus.
See android.lens.info.focusDistanceCalibration for details
>= 0
If the lens is fixed-focus, this will be
0.
Mandatory for FULL devices; LIMITED devices
must always set this value to 0 for fixed-focus; and may omit
the minimum focus distance otherwise.
This field is also mandatory for all devices advertising
the MANUAL_SENSOR capability.
2
Dimensions of lens shading map.
Both values >= 1
The map should be on the order of 30-40 rows and columns, and
must be smaller than 64x64.
UNCALIBRATED
The lens focus distance is not accurate, and the units used for
android.lens.focusDistance do not correspond to any physical units.
Setting the lens to the same focus distance on separate occasions may
result in a different real focus distance, depending on factors such
as the orientation of the device, the age of the focusing mechanism,
and the device temperature. The focus distance value will still be
in the range of `[0, android.lens.info.minimumFocusDistance]`, where 0
represents the farthest focus.
APPROXIMATE
The lens focus distance is measured in diopters.
However, setting the lens to the same focus distance
on separate occasions may result in a different real
focus distance, depending on factors such as the
orientation of the device, the age of the focusing
mechanism, and the device temperature.
CALIBRATED
The lens focus distance is measured in diopters, and
is calibrated.
The lens mechanism is calibrated so that setting the
same focus distance is repeatable on multiple
occasions with good accuracy, and the focus distance
corresponds to the real physical distance to the plane
of best focus.
The lens focus distance calibration quality.
The lens focus distance calibration quality determines the reliability of
focus related metadata entries, i.e. android.lens.focusDistance,
android.lens.focusRange, android.lens.info.hyperfocalDistance, and
android.lens.info.minimumFocusDistance.
APPROXIMATE and CALIBRATED devices report the focus metadata in
units of diopters (1/meter), so `0.0f` represents focusing at infinity,
and increasing positive numbers represent focusing closer and closer
to the camera device. The focus distance control also uses diopters
on these devices.
UNCALIBRATED devices do not use units that are directly comparable
to any real physical measurement, but `0.0f` still represents farthest
focus, and android.lens.info.minimumFocusDistance represents the
nearest focus the device can achieve.
For devices advertise APPROXIMATE quality or higher, diopters 0 (infinity
focus) must work. When autofocus is disabled (android.control.afMode == OFF)
and the lens focus distance is set to 0 diopters
(android.lens.focusDistance == 0), the lens will move to focus at infinity
and is stably focused at infinity even if the device tilts. It may take the
lens some time to move; during the move the lens state should be MOVING and
the output diopter value should be changing toward 0.
FRONT
The camera device faces the same direction as the device's screen.
BACK
The camera device faces the opposite direction as the device's screen.
EXTERNAL
The camera device is an external camera, and has no fixed facing relative to the
device's screen.
Direction the camera faces relative to
device screen.
4
The orientation of the camera relative to the sensor
coordinate system.
Quaternion coefficients
The four coefficients that describe the quaternion
rotation from the Android sensor coordinate system to a
camera-aligned coordinate system where the X-axis is
aligned with the long side of the image sensor, the Y-axis
is aligned with the short side of the image sensor, and
the Z-axis is aligned with the optical axis of the sensor.
To convert from the quaternion coefficients `(x,y,z,w)`
to the axis of rotation `(a_x, a_y, a_z)` and rotation
amount `theta`, the following formulas can be used:
theta = 2 * acos(w)
a_x = x / sin(theta/2)
a_y = y / sin(theta/2)
a_z = z / sin(theta/2)
To create a 3x3 rotation matrix that applies the rotation
defined by this quaternion, the following matrix can be
used:
R = [ 1 - 2y^2 - 2z^2, 2xy - 2zw, 2xz + 2yw,
2xy + 2zw, 1 - 2x^2 - 2z^2, 2yz - 2xw,
2xz - 2yw, 2yz + 2xw, 1 - 2x^2 - 2y^2 ]
This matrix can then be used to apply the rotation to a
column vector point with
`p' = Rp`
where `p` is in the device sensor coordinate system, and
`p'` is in the camera-oriented coordinate system.
3
Position of the camera optical center.
Meters
The position of the camera device's lens optical center,
as a three-dimensional vector `(x,y,z)`.
Prior to Android P, or when android.lens.poseReference is PRIMARY_CAMERA, this position
is relative to the optical center of the largest camera device facing in the same
direction as this camera, in the {@link android.hardware.SensorEvent Android sensor
coordinate axes}. Note that only the axis definitions are shared with the sensor
coordinate system, but not the origin.
If this device is the largest or only camera device with a given facing, then this
position will be `(0, 0, 0)`; a camera device with a lens optical center located 3 cm
from the main sensor along the +X axis (to the right from the user's perspective) will
report `(0.03, 0, 0)`. Note that this means that, for many computer vision
applications, the position needs to be negated to convert it to a translation from the
camera to the origin.
To transform a pixel coordinates between two cameras facing the same direction, first
the source camera android.lens.distortion must be corrected for. Then the source
camera android.lens.intrinsicCalibration needs to be applied, followed by the
android.lens.poseRotation of the source camera, the translation of the source camera
relative to the destination camera, the android.lens.poseRotation of the destination
camera, and finally the inverse of android.lens.intrinsicCalibration of the destination
camera. This obtains a radial-distortion-free coordinate in the destination camera pixel
coordinates.
To compare this against a real image from the destination camera, the destination camera
image then needs to be corrected for radial distortion before comparison or sampling.
When android.lens.poseReference is GYROSCOPE, then this position is relative to
the center of the primary gyroscope on the device. The axis definitions are the same as
with PRIMARY_CAMERA.
Should be zero for fixed-focus cameras
2
The range of scene distances that are in
sharp focus (depth of field).
A pair of focus distances in diopters: (near,
far); see android.lens.info.focusDistanceCalibration for details.
>=0
If variable focus not supported, can still report
fixed depth of field range
STATIONARY
The lens parameters (android.lens.focalLength, android.lens.focusDistance,
android.lens.filterDensity and android.lens.aperture) are not changing.
MOVING
One or several of the lens parameters
(android.lens.focalLength, android.lens.focusDistance,
android.lens.filterDensity or android.lens.aperture) is
currently changing.
Current lens status.
For lens parameters android.lens.focalLength, android.lens.focusDistance,
android.lens.filterDensity and android.lens.aperture, when changes are requested,
they may take several frames to reach the requested values. This state indicates
the current status of the lens parameters.
When the state is STATIONARY, the lens parameters are not changing. This could be
either because the parameters are all fixed, or because the lens has had enough
time to reach the most recently-requested values.
If all these lens parameters are not changable for a camera device, as listed below:
* Fixed focus (`android.lens.info.minimumFocusDistance == 0`), which means
android.lens.focusDistance parameter will always be 0.
* Fixed focal length (android.lens.info.availableFocalLengths contains single value),
which means the optical zoom is not supported.
* No ND filter (android.lens.info.availableFilterDensities contains only 0).
* Fixed aperture (android.lens.info.availableApertures contains single value).
Then this state will always be STATIONARY.
When the state is MOVING, it indicates that at least one of the lens parameters
is changing.
5
The parameters for this camera device's intrinsic
calibration.
Pixels in the
android.sensor.info.preCorrectionActiveArraySize
coordinate system.
The five calibration parameters that describe the
transform from camera-centric 3D coordinates to sensor
pixel coordinates:
[f_x, f_y, c_x, c_y, s]
Where `f_x` and `f_y` are the horizontal and vertical
focal lengths, `[c_x, c_y]` is the position of the optical
axis, and `s` is a skew parameter for the sensor plane not
being aligned with the lens plane.
These are typically used within a transformation matrix K:
K = [ f_x, s, c_x,
0, f_y, c_y,
0 0, 1 ]
which can then be combined with the camera pose rotation
`R` and translation `t` (android.lens.poseRotation and
android.lens.poseTranslation, respectively) to calculate the
complete transform from world coordinates to pixel
coordinates:
P = [ K 0 * [ R -Rt
0 1 ] 0 1 ]
(Note the negation of poseTranslation when mapping from camera
to world coordinates, and multiplication by the rotation).
With `p_w` being a point in the world coordinate system
and `p_s` being a point in the camera active pixel array
coordinate system, and with the mapping including the
homogeneous division by z:
p_h = (x_h, y_h, z_h) = P p_w
p_s = p_h / z_h
so `[x_s, y_s]` is the pixel coordinates of the world
point, `z_s = 1`, and `w_s` is a measurement of disparity
(depth) in pixel coordinates.
Note that the coordinate system for this transform is the
android.sensor.info.preCorrectionActiveArraySize system,
where `(0,0)` is the top-left of the
preCorrectionActiveArraySize rectangle. Once the pose and
intrinsic calibration transforms have been applied to a
world point, then the android.lens.distortion
transform needs to be applied, and the result adjusted to
be in the android.sensor.info.activeArraySize coordinate
system (where `(0, 0)` is the top-left of the
activeArraySize rectangle), to determine the final pixel
coordinate of the world point for processed (non-RAW)
output buffers.
For camera devices, the center of pixel `(x,y)` is located at
coordinate `(x + 0.5, y + 0.5)`. So on a device with a
precorrection active array of size `(10,10)`, the valid pixel
indices go from `(0,0)-(9,9)`, and an perfectly-built camera would
have an optical center at the exact center of the pixel grid, at
coordinates `(5.0, 5.0)`, which is the top-left corner of pixel
`(5,5)`.
6
The correction coefficients to correct for this camera device's
radial and tangential lens distortion.
This field was inconsistently defined in terms of its
normalization. Use android.lens.distortion instead.
Unitless coefficients.
Four radial distortion coefficients `[kappa_0, kappa_1, kappa_2,
kappa_3]` and two tangential distortion coefficients
`[kappa_4, kappa_5]` that can be used to correct the
lens's geometric distortion with the mapping equations:
x_c = x_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
kappa_4 * (2 * x_i * y_i) + kappa_5 * ( r^2 + 2 * x_i^2 )
y_c = y_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
kappa_5 * (2 * x_i * y_i) + kappa_4 * ( r^2 + 2 * y_i^2 )
Here, `[x_c, y_c]` are the coordinates to sample in the
input image that correspond to the pixel values in the
corrected image at the coordinate `[x_i, y_i]`:
correctedImage(x_i, y_i) = sample_at(x_c, y_c, inputImage)
The pixel coordinates are defined in a normalized
coordinate system related to the
android.lens.intrinsicCalibration calibration fields.
Both `[x_i, y_i]` and `[x_c, y_c]` have `(0,0)` at the
lens optical center `[c_x, c_y]`. The maximum magnitudes
of both x and y coordinates are normalized to be 1 at the
edge further from the optical center, so the range
for both dimensions is `-1 <= x <= 1`.
Finally, `r` represents the radial distance from the
optical center, `r^2 = x_i^2 + y_i^2`, and its magnitude
is therefore no larger than `|r| <= sqrt(2)`.
The distortion model used is the Brown-Conrady model.
PRIMARY_CAMERA
The value of android.lens.poseTranslation is relative to the optical center of
the largest camera device facing the same direction as this camera.
This is the default value for API levels before Android P.
GYROSCOPE
The value of android.lens.poseTranslation is relative to the position of the
primary gyroscope of this Android device.
The origin for android.lens.poseTranslation.
Different calibration methods and use cases can produce better or worse results
depending on the selected coordinate origin.
5
The correction coefficients to correct for this camera device's
radial and tangential lens distortion.
Replaces the deprecated android.lens.radialDistortion field, which was
inconsistently defined.
Unitless coefficients.
Three radial distortion coefficients `[kappa_1, kappa_2,
kappa_3]` and two tangential distortion coefficients
`[kappa_4, kappa_5]` that can be used to correct the
lens's geometric distortion with the mapping equations:
x_c = x_i * ( 1 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
kappa_4 * (2 * x_i * y_i) + kappa_5 * ( r^2 + 2 * x_i^2 )
y_c = y_i * ( 1 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) +
kappa_5 * (2 * x_i * y_i) + kappa_4 * ( r^2 + 2 * y_i^2 )
Here, `[x_c, y_c]` are the coordinates to sample in the
input image that correspond to the pixel values in the
corrected image at the coordinate `[x_i, y_i]`:
correctedImage(x_i, y_i) = sample_at(x_c, y_c, inputImage)
The pixel coordinates are defined in a coordinate system
related to the android.lens.intrinsicCalibration
calibration fields; see that entry for details of the mapping stages.
Both `[x_i, y_i]` and `[x_c, y_c]`
have `(0,0)` at the lens optical center `[c_x, c_y]`, and
the range of the coordinates depends on the focal length
terms of the intrinsic calibration.
Finally, `r` represents the radial distance from the
optical center, `r^2 = x_i^2 + y_i^2`.
The distortion model used is the Brown-Conrady model.
OFF
No noise reduction is applied.
FAST
Noise reduction is applied without reducing frame rate relative to sensor
output. It may be the same as OFF if noise reduction will reduce frame rate
relative to sensor.
HIGH_QUALITY
High-quality noise reduction is applied, at the cost of possibly reduced frame
rate relative to sensor output.
MINIMAL
MINIMAL noise reduction is applied without reducing frame rate relative to
sensor output.
ZERO_SHUTTER_LAG
Noise reduction is applied at different levels for different output streams,
based on resolution. Streams at maximum recording resolution (see {@link
android.hardware.camera2.CameraDevice#createCaptureSession|ACameraDevice_createCaptureSession})
or below have noise reduction applied, while higher-resolution streams have MINIMAL (if
supported) or no noise reduction applied (if MINIMAL is not supported.) The degree of
noise reduction for low-resolution streams is tuned so that frame rate is not impacted,
and the quality is equal to or better than FAST (since it is only applied to
lower-resolution outputs, quality may improve from FAST).
This mode is intended to be used by applications operating in a zero-shutter-lag mode
with YUV or PRIVATE reprocessing, where the application continuously captures
high-resolution intermediate buffers into a circular buffer, from which a final image is
produced via reprocessing when a user takes a picture. For such a use case, the
high-resolution buffers must not have noise reduction applied to maximize efficiency of
preview and to avoid over-applying noise filtering when reprocessing, while
low-resolution buffers (used for recording or preview, generally) need noise reduction
applied for reasonable preview quality.
This mode is guaranteed to be supported by devices that support either the
YUV_REPROCESSING or PRIVATE_REPROCESSING capabilities
(android.request.availableCapabilities lists either of those capabilities) and it will
be the default mode for CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template.
Mode of operation for the noise reduction algorithm.
android.noiseReduction.availableNoiseReductionModes
The noise reduction algorithm attempts to improve image quality by removing
excessive noise added by the capture process, especially in dark conditions.
OFF means no noise reduction will be applied by the camera device, for both raw and
YUV domain.
MINIMAL means that only sensor raw domain basic noise reduction is enabled ,to remove
demosaicing or other processing artifacts. For YUV_REPROCESSING, MINIMAL is same as OFF.
This mode is optional, may not be support by all devices. The application should check
android.noiseReduction.availableNoiseReductionModes before using it.
FAST/HIGH_QUALITY both mean camera device determined noise filtering
will be applied. HIGH_QUALITY mode indicates that the camera device
will use the highest-quality noise filtering algorithms,
even if it slows down capture rate. FAST means the camera device will not
slow down capture rate when applying noise filtering. FAST may be the same as MINIMAL if
MINIMAL is listed, or the same as OFF if any noise filtering will slow down capture rate.
Every output stream will have a similar amount of enhancement applied.
ZERO_SHUTTER_LAG is meant to be used by applications that maintain a continuous circular
buffer of high-resolution images during preview and reprocess image(s) from that buffer
into a final capture when triggered by the user. In this mode, the camera device applies
noise reduction to low-resolution streams (below maximum recording resolution) to maximize
preview quality, but does not apply noise reduction to high-resolution streams, since
those will be reprocessed later if necessary.
For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device
will apply FAST/HIGH_QUALITY YUV domain noise reduction, respectively. The camera device
may adjust the noise reduction parameters for best image quality based on the
android.reprocess.effectiveExposureFactor if it is set.
For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to
adjust the internal noise reduction parameters appropriately to get the best quality
images.
Control the amount of noise reduction
applied to the images
1-10; 10 is max noise reduction
1 - 10
n
List of noise reduction modes for android.noiseReduction.mode that are supported
by this camera device.
Any value listed in android.noiseReduction.mode
Full-capability camera devices will always support OFF and FAST.
Camera devices that support YUV_REPROCESSING or PRIVATE_REPROCESSING will support
ZERO_SHUTTER_LAG.
Legacy-capability camera devices will only support FAST mode.
HAL must support both FAST and HIGH_QUALITY if noise reduction control is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY will generate the same output.
If set to 1, the camera service does not
scale 'normalized' coordinates with respect to the crop
region. This applies to metering input (a{e,f,wb}Region
and output (face rectangles).
Not used in HALv3 or newer
Normalized coordinates refer to those in the
(-1000,1000) range mentioned in the
android.hardware.Camera API.
HAL implementations should instead always use and emit
sensor array-relative coordinates for all region data. Does
not need to be listed in static metadata. Support will be
removed in future versions of camera service.
If set to 1, then the camera service always
switches to FOCUS_MODE_AUTO before issuing a AF
trigger.
Not used in HALv3 or newer
HAL implementations should implement AF trigger
modes for AUTO, MACRO, CONTINUOUS_FOCUS, and
CONTINUOUS_PICTURE modes instead of using this flag. Does
not need to be listed in static metadata. Support will be
removed in future versions of camera service
If set to 1, the camera service uses
CAMERA2_PIXEL_FORMAT_ZSL instead of
HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED for the zero
shutter lag stream
Not used in HALv3 or newer
HAL implementations should use gralloc usage flags
to determine that a stream will be used for
zero-shutter-lag, instead of relying on an explicit
format setting. Does not need to be listed in static
metadata. Support will be removed in future versions of
camera service.
If set to 1, the HAL will always split result
metadata for a single capture into multiple buffers,
returned using multiple process_capture_result calls.
Not used in HALv3 or newer; replaced by better partials mechanism
Does not need to be listed in static
metadata. Support for partial results will be reworked in
future versions of camera service. This quirk will stop
working at that point; DO NOT USE without careful
consideration of future support.
Refer to `camera3_capture_result::partial_result`
for information on how to implement partial results.
FINAL
The last or only metadata result buffer
for this capture.
PARTIAL
A partial buffer of result metadata for this
capture. More result buffers for this capture will be sent
by the camera device, the last of which will be marked
FINAL.
Whether a result given to the framework is the
final one for the capture, or only a partial that contains a
subset of the full set of dynamic metadata
values.
Not used in HALv3 or newer
Optional. Default value is FINAL.
The entries in the result metadata buffers for a
single capture may not overlap, except for this entry. The
FINAL buffers must retain FIFO ordering relative to the
requests that generate them, so the FINAL buffer for frame 3 must
always be sent to the framework after the FINAL buffer for frame 2, and
before the FINAL buffer for frame 4. PARTIAL buffers may be returned
in any order relative to other frames, but all PARTIAL buffers for a given
capture must arrive before the FINAL buffer for that capture. This entry may
only be used by the camera device if quirks.usePartialResult is set to 1.
Refer to `camera3_capture_result::partial_result`
for information on how to implement partial results.
A frame counter set by the framework. Must
be maintained unchanged in output frame. This value monotonically
increases with every new result (that is, each new result has a unique
frameCount value).
Not used in HALv3 or newer
incrementing integer
Any int.
An application-specified ID for the current
request. Must be maintained unchanged in output
frame
arbitrary integer assigned by application
Any int
n
List which camera reprocess stream is used
for the source of reprocessing data.
Not used in HALv3 or newer
List of camera reprocess stream IDs
Typically, only one entry allowed, must be a valid reprocess stream ID.
Only meaningful when android.request.type ==
REPROCESS. Ignored otherwise
NONE
No metadata should be produced on output, except
for application-bound buffer data. If no
application-bound streams exist, no frame should be
placed in the output frame queue. If such streams
exist, a frame should be placed on the output queue
with null metadata but with the necessary output buffer
information. Timestamp information should still be
included with any output stream buffers
FULL
All metadata should be produced. Statistics will
only be produced if they are separately
enabled
How much metadata to produce on
output
n
Lists which camera output streams image data
from this capture must be sent to
Not used in HALv3 or newer
List of camera stream IDs
List must only include streams that have been
created
If no output streams are listed, then the image
data should simply be discarded. The image data must
still be captured for metadata and statistics production,
and the lens and flash must operate as requested.
CAPTURE
Capture a new image from the imaging hardware,
and process it according to the
settings
REPROCESS
Process previously captured data; the
android.request.inputStreams parameter determines the
source reprocessing stream. TODO: Mark dynamic metadata
needed for reprocessing with [RP]
The type of the request; either CAPTURE or
REPROCESS. For legacy HAL3, this tag is redundant.
Not used in HALv3 or newer
3
The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device.
For processed (and stalling) format streams, >= 1.
For Raw format (either stalling or non-stalling) streams, >= 0.
For processed (but not stalling) format streams, >= 3
for FULL mode devices (`android.info.supportedHardwareLevel == FULL`);
>= 2 for LIMITED mode devices (`android.info.supportedHardwareLevel == LIMITED`).
This is a 3 element tuple that contains the max number of output simultaneous
streams for raw sensor, processed (but not stalling), and processed (and stalling)
formats respectively. For example, assuming that JPEG is typically a processed and
stalling stream, if max raw sensor format output stream number is 1, max YUV streams
number is 3, and max JPEG stream number is 2, then this tuple should be `(1, 3, 2)`.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for an output stream can
be any supported format provided by android.scaler.availableStreamConfigurations.
The formats defined in android.scaler.availableStreamConfigurations can be catergorized
into the 3 stream types as below:
* Processed (but stalling): any non-RAW format with a stallDurations > 0.
Typically {@link android.graphics.ImageFormat#JPEG|AIMAGE_FORMAT_JPEG JPEG format}.
* Raw formats: {@link android.graphics.ImageFormat#RAW_SENSOR|AIMAGE_FORMAT_RAW16
RAW_SENSOR}, {@link android.graphics.ImageFormat#RAW10|AIMAGE_FORMAT_RAW10 RAW10}, or
{@link android.graphics.ImageFormat#RAW12|AIMAGE_FORMAT_RAW12 RAW12}.
* Processed (but not-stalling): any non-RAW format without a stall duration. Typically
{@link android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888 YUV_420_888},
{@link android.graphics.ImageFormat#NV21 NV21}, {@link
android.graphics.ImageFormat#YV12 YV12}, or {@link
android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8 Y8} .
The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device
for any `RAW` formats.
>= 0
This value contains the max number of output simultaneous
streams from the raw sensor.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for this kind of an output stream can
be any `RAW` and supported format provided by android.scaler.streamConfigurationMap.
In particular, a `RAW` format is typically one of:
* {@link android.graphics.ImageFormat#RAW_SENSOR|AIMAGE_FORMAT_RAW16 RAW_SENSOR}
* {@link android.graphics.ImageFormat#RAW10|AIMAGE_FORMAT_RAW10 RAW10}
* {@link android.graphics.ImageFormat#RAW12|AIMAGE_FORMAT_RAW12 RAW12}
LEGACY mode devices (android.info.supportedHardwareLevel `==` LEGACY)
never support raw streams.
The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device
for any processed (but not-stalling) formats.
>= 3
for FULL mode devices (`android.info.supportedHardwareLevel == FULL`);
>= 2 for LIMITED mode devices (`android.info.supportedHardwareLevel == LIMITED`).
This value contains the max number of output simultaneous
streams for any processed (but not-stalling) formats.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for this kind of an output stream can
be any non-`RAW` and supported format provided by android.scaler.streamConfigurationMap.
Processed (but not-stalling) is defined as any non-RAW format without a stall duration.
Typically:
* {@link android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888 YUV_420_888}
* {@link android.graphics.ImageFormat#NV21 NV21}
* {@link android.graphics.ImageFormat#YV12 YV12}
* Implementation-defined formats, i.e. {@link
android.hardware.camera2.params.StreamConfigurationMap#isOutputSupportedFor(Class)}
* {@link android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8 Y8}
For full guarantees, query {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration} with a
processed format -- it will return 0 for a non-stalling stream.
LEGACY devices will support at least 2 processing/non-stalling streams.
The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device
for any processed (and stalling) formats.
>= 1
This value contains the max number of output simultaneous
streams for any processed (but not-stalling) formats.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for this kind of an output stream can
be any non-`RAW` and supported format provided by android.scaler.streamConfigurationMap.
A processed and stalling format is defined as any non-RAW format with a stallDurations
> 0. Typically only the {@link
android.graphics.ImageFormat#JPEG|AIMAGE_FORMAT_JPEG JPEG format} is a stalling format.
For full guarantees, query {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration} with a
processed format -- it will return a non-0 value for a stalling stream.
LEGACY devices will support up to 1 processing/stalling stream.
1
How many reprocessing streams of any type
can be allocated at the same time.
Not used in HALv3 or newer
>= 0
Only used by HAL2.x.
When set to 0, it means no reprocess stream is supported.
The maximum numbers of any type of input streams
that can be configured and used simultaneously by a camera device.
0 or 1.
When set to 0, it means no input stream is supported.
The image format for a input stream can be any supported format returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputFormats}. When using an
input stream, there must be at least one output stream configured to to receive the
reprocessed images.
When an input stream and some output streams are used in a reprocessing request,
only the input buffer will be used to produce these output stream buffers, and a
new sensor image will not be captured.
For example, for Zero Shutter Lag (ZSL) still capture use case, the input
stream image format will be PRIVATE, the associated output stream image format
should be JPEG.
For the reprocessing flow and controls, see
hardware/libhardware/include/hardware/camera3.h Section 10 for more details.
A frame counter set by the framework. This value monotonically
increases with every new result (that is, each new result has a unique
frameCount value).
Not used in HALv3 or newer
count of frames
> 0
Reset on release()
Specifies the number of pipeline stages the frame went
through from when it was exposed to when the final completed result
was available to the framework.
<= android.request.pipelineMaxDepth
Depending on what settings are used in the request, and
what streams are configured, the data may undergo less processing,
and some pipeline stages skipped.
See android.request.pipelineMaxDepth for more details.
This value must always represent the accurate count of how many
pipeline stages were actually used.
Specifies the number of maximum pipeline stages a frame
has to go through from when it's exposed to when it's available
to the framework.
A typical minimum value for this is 2 (one stage to expose,
one stage to readout) from the sensor. The ISP then usually adds
its own stages to do custom HW processing. Further stages may be
added by SW processing.
Depending on what settings are used (e.g. YUV, JPEG) and what
processing is enabled (e.g. face detection), the actual pipeline
depth (specified by android.request.pipelineDepth) may be less than
the max pipeline depth.
A pipeline depth of X stages is equivalent to a pipeline latency of
X frame intervals.
This value will normally be 8 or less, however, for high speed capture session,
the max pipeline depth will be up to 8 x size of high speed capture request list.
This value should be 4 or less, expect for the high speed recording session, where the
max batch sizes may be larger than 1.
Defines how many sub-components
a result will be composed of.
>= 1
In order to combat the pipeline latency, partial results
may be delivered to the application layer from the camera device as
soon as they are available.
Optional; defaults to 1. A value of 1 means that partial
results are not supported, and only the final TotalCaptureResult will
be produced by the camera device.
A typical use case for this might be: after requesting an
auto-focus (AF) lock the new AF state might be available 50%
of the way through the pipeline. The camera device could
then immediately dispatch this state via a partial result to
the application, and the rest of the metadata via later
partial results.
n
BACKWARD_COMPATIBLE
The minimal set of capabilities that every camera
device (regardless of android.info.supportedHardwareLevel)
supports.
This capability is listed by all normal devices, and
indicates that the camera device has a feature set
that's comparable to the baseline requirements for the
older android.hardware.Camera API.
Devices with the DEPTH_OUTPUT capability might not list this
capability, indicating that they support only depth measurement,
not standard color output.
MANUAL_SENSOR
The camera device can be manually controlled (3A algorithms such
as auto-exposure, and auto-focus can be bypassed).
The camera device supports basic manual control of the sensor image
acquisition related stages. This means the following controls are
guaranteed to be supported:
* Manual frame duration control
* android.sensor.frameDuration
* android.sensor.info.maxFrameDuration
* Manual exposure control
* android.sensor.exposureTime
* android.sensor.info.exposureTimeRange
* Manual sensitivity control
* android.sensor.sensitivity
* android.sensor.info.sensitivityRange
* Manual lens control (if the lens is adjustable)
* android.lens.*
* Manual flash control (if a flash unit is present)
* android.flash.*
* Manual black level locking
* android.blackLevel.lock
* Auto exposure lock
* android.control.aeLock
If any of the above 3A algorithms are enabled, then the camera
device will accurately report the values applied by 3A in the
result.
A given camera device may also support additional manual sensor controls,
but this capability only covers the above list of controls.
If this is supported, android.scaler.streamConfigurationMap will
additionally return a min frame duration that is greater than
zero for each supported size-format combination.
For camera devices with LOGICAL_MULTI_CAMERA capability, when the underlying active
physical camera switches, exposureTime, sensitivity, and lens properties may change
even if AE/AF is locked. However, the overall auto exposure and auto focus experience
for users will be consistent. Refer to LOGICAL_MULTI_CAMERA capability for details.
MANUAL_POST_PROCESSING
The camera device post-processing stages can be manually controlled.
The camera device supports basic manual control of the image post-processing
stages. This means the following controls are guaranteed to be supported:
* Manual tonemap control
* android.tonemap.curve
* android.tonemap.mode
* android.tonemap.maxCurvePoints
* android.tonemap.gamma
* android.tonemap.presetCurve
* Manual white balance control
* android.colorCorrection.transform
* android.colorCorrection.gains
* Manual lens shading map control
* android.shading.mode
* android.statistics.lensShadingMapMode
* android.statistics.lensShadingMap
* android.lens.info.shadingMapSize
* Manual aberration correction control (if aberration correction is supported)
* android.colorCorrection.aberrationMode
* android.colorCorrection.availableAberrationModes
* Auto white balance lock
* android.control.awbLock
If auto white balance is enabled, then the camera device
will accurately report the values applied by AWB in the result.
A given camera device may also support additional post-processing
controls, but this capability only covers the above list of controls.
For camera devices with LOGICAL_MULTI_CAMERA capability, when underlying active
physical camera switches, tonemap, white balance, and shading map may change even if
awb is locked. However, the overall post-processing experience for users will be
consistent. Refer to LOGICAL_MULTI_CAMERA capability for details.
RAW
The camera device supports outputting RAW buffers and
metadata for interpreting them.
Devices supporting the RAW capability allow both for
saving DNG files, and for direct application processing of
raw sensor images.
* RAW_SENSOR is supported as an output format.
* The maximum available resolution for RAW_SENSOR streams
will match either the value in
android.sensor.info.pixelArraySize or
android.sensor.info.preCorrectionActiveArraySize.
* All DNG-related optional metadata entries are provided
by the camera device.
PRIVATE_REPROCESSING
The camera device supports the Zero Shutter Lag reprocessing use case.
* One input stream is supported, that is, `android.request.maxNumInputStreams == 1`.
* {@link android.graphics.ImageFormat#PRIVATE} is supported as an output/input format,
that is, {@link android.graphics.ImageFormat#PRIVATE} is included in the lists of
formats returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputFormats} and {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputFormats}.
* {@link android.hardware.camera2.params.StreamConfigurationMap#getValidOutputFormatsForInput}
returns non empty int[] for each supported input format returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputFormats}.
* Each size returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputSizes
getInputSizes(ImageFormat.PRIVATE)} is also included in {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes
getOutputSizes(ImageFormat.PRIVATE)}
* Using {@link android.graphics.ImageFormat#PRIVATE} does not cause a frame rate drop
relative to the sensor's maximum capture rate (at that resolution).
* {@link android.graphics.ImageFormat#PRIVATE} will be reprocessable into both
{@link android.graphics.ImageFormat#YUV_420_888} and
{@link android.graphics.ImageFormat#JPEG} formats.
* For a MONOCHROME camera supporting Y8 format, {@link
android.graphics.ImageFormat#PRIVATE} will be reprocessable into
{@link android.graphics.ImageFormat#Y8}.
* The maximum available resolution for PRIVATE streams
(both input/output) will match the maximum available
resolution of JPEG streams.
* Static metadata android.reprocess.maxCaptureStall.
* Only below controls are effective for reprocessing requests and
will be present in capture results, other controls in reprocess
requests will be ignored by the camera device.
* android.jpeg.*
* android.noiseReduction.mode
* android.edge.mode
* android.noiseReduction.availableNoiseReductionModes and
android.edge.availableEdgeModes will both list ZERO_SHUTTER_LAG as a supported mode.
READ_SENSOR_SETTINGS
The camera device supports accurately reporting the sensor settings for many of
the sensor controls while the built-in 3A algorithm is running. This allows
reporting of sensor settings even when these settings cannot be manually changed.
The values reported for the following controls are guaranteed to be available
in the CaptureResult, including when 3A is enabled:
* Exposure control
* android.sensor.exposureTime
* Sensitivity control
* android.sensor.sensitivity
* Lens controls (if the lens is adjustable)
* android.lens.focusDistance
* android.lens.aperture
This capability is a subset of the MANUAL_SENSOR control capability, and will
always be included if the MANUAL_SENSOR capability is available.
BURST_CAPTURE
The camera device supports capturing high-resolution images at >= 20 frames per
second, in at least the uncompressed YUV format, when post-processing settings are
set to FAST. Additionally, all image resolutions less than 24 megapixels can be
captured at >= 10 frames per second. Here, 'high resolution' means at least 8
megapixels, or the maximum resolution of the device, whichever is smaller.
More specifically, this means that a size matching the camera device's active array
size is listed as a supported size for the {@link
android.graphics.ImageFormat#YUV_420_888} format in either {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes} or {@link
android.hardware.camera2.params.StreamConfigurationMap#getHighResolutionOutputSizes},
with a minimum frame duration for that format and size of either <= 1/20 s, or
<= 1/10 s if the image size is less than 24 megapixels, respectively; and
the android.control.aeAvailableTargetFpsRanges entry lists at least one FPS range
where the minimum FPS is >= 1 / minimumFrameDuration for the maximum-size
YUV_420_888 format. If that maximum size is listed in {@link
android.hardware.camera2.params.StreamConfigurationMap#getHighResolutionOutputSizes},
then the list of resolutions for YUV_420_888 from {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes} contains at
least one resolution >= 8 megapixels, with a minimum frame duration of <= 1/20
s.
If the device supports the {@link
android.graphics.ImageFormat#RAW10|AIMAGE_FORMAT_RAW10}, {@link
android.graphics.ImageFormat#RAW12|AIMAGE_FORMAT_RAW12}, {@link
android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8}, then those can also be
captured at the same rate as the maximum-size YUV_420_888 resolution is.
If the device supports the PRIVATE_REPROCESSING capability, then the same guarantees
as for the YUV_420_888 format also apply to the {@link
android.graphics.ImageFormat#PRIVATE} format.
In addition, the android.sync.maxLatency field is guaranted to have a value between 0
and 4, inclusive. android.control.aeLockAvailable and android.control.awbLockAvailable
are also guaranteed to be `true` so burst capture with these two locks ON yields
consistent image output.
More specifically, this means that at least one output {@link
android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888} size listed in
{@link
android.hardware.camera2.params.StreamConfigurationMap|ACAMERA_SCALER_AVAILABLE_STREAM_CONFIGURATIONS}
is larger or equal to the 'high resolution' defined above, and can be captured at at
least 20 fps. For the largest {@link
android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888} size listed in
{@link
android.hardware.camera2.params.StreamConfigurationMap|ACAMERA_SCALER_AVAILABLE_STREAM_CONFIGURATIONS},
camera device can capture this size for at least 10 frames per second if the size is
less than 24 megapixels. Also the android.control.aeAvailableTargetFpsRanges entry
lists at least one FPS range where the minimum FPS is >= 1 / minimumFrameDuration
for the largest YUV_420_888 size.
If the device supports the {@link
android.graphics.ImageFormat#RAW10|AIMAGE_FORMAT_RAW10}, {@link
android.graphics.ImageFormat#RAW12|AIMAGE_FORMAT_RAW12}, {@link
android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8}, then those can also be
captured at the same rate as the maximum-size YUV_420_888 resolution is.
In addition, the android.sync.maxLatency field is guaranted to have a value between 0
and 4, inclusive. android.control.aeLockAvailable and android.control.awbLockAvailable
are also guaranteed to be `true` so burst capture with these two locks ON yields
consistent image output.
YUV_REPROCESSING
The camera device supports the YUV_420_888 reprocessing use case, similar as
PRIVATE_REPROCESSING, This capability requires the camera device to support the
following:
* One input stream is supported, that is, `android.request.maxNumInputStreams == 1`.
* {@link android.graphics.ImageFormat#YUV_420_888} is supported as an output/input
format, that is, YUV_420_888 is included in the lists of formats returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputFormats} and {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputFormats}.
* {@link
android.hardware.camera2.params.StreamConfigurationMap#getValidOutputFormatsForInput}
returns non-empty int[] for each supported input format returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputFormats}.
* Each size returned by {@link
android.hardware.camera2.params.StreamConfigurationMap#getInputSizes
getInputSizes(YUV_420_888)} is also included in {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes
getOutputSizes(YUV_420_888)}
* Using {@link android.graphics.ImageFormat#YUV_420_888} does not cause a frame rate
drop relative to the sensor's maximum capture rate (at that resolution).
* {@link android.graphics.ImageFormat#YUV_420_888} will be reprocessable into both
{@link android.graphics.ImageFormat#YUV_420_888} and {@link
android.graphics.ImageFormat#JPEG} formats.
* The maximum available resolution for {@link
android.graphics.ImageFormat#YUV_420_888} streams (both input/output) will match the
maximum available resolution of {@link android.graphics.ImageFormat#JPEG} streams.
* For a MONOCHROME camera with Y8 format support, all the requirements mentioned
above for YUV_420_888 apply for Y8 format as well.
* Static metadata android.reprocess.maxCaptureStall.
* Only the below controls are effective for reprocessing requests and will be present
in capture results. The reprocess requests are from the original capture results
that are associated with the intermediate {@link
android.graphics.ImageFormat#YUV_420_888} output buffers. All other controls in the
reprocess requests will be ignored by the camera device.
* android.jpeg.*
* android.noiseReduction.mode
* android.edge.mode
* android.reprocess.effectiveExposureFactor
* android.noiseReduction.availableNoiseReductionModes and
android.edge.availableEdgeModes will both list ZERO_SHUTTER_LAG as a supported mode.
DEPTH_OUTPUT
The camera device can produce depth measurements from its field of view.
This capability requires the camera device to support the following:
* {@link android.graphics.ImageFormat#DEPTH16|AIMAGE_FORMAT_DEPTH16} is supported as
an output format.
* {@link
android.graphics.ImageFormat#DEPTH_POINT_CLOUD|AIMAGE_FORMAT_DEPTH_POINT_CLOUD} is
optionally supported as an output format.
* This camera device, and all camera devices with the same android.lens.facing, will
list the following calibration metadata entries in both {@link
android.hardware.camera2.CameraCharacteristics|ACameraManager_getCameraCharacteristics}
and {@link
android.hardware.camera2.CaptureResult|ACameraCaptureSession_captureCallback_result}:
- android.lens.poseTranslation
- android.lens.poseRotation
- android.lens.intrinsicCalibration
- android.lens.distortion
* The android.depth.depthIsExclusive entry is listed by this device.
* As of Android P, the android.lens.poseReference entry is listed by this device.
* A LIMITED camera with only the DEPTH_OUTPUT capability does not have to support
normal YUV_420_888, Y8, JPEG, and PRIV-format outputs. It only has to support the
DEPTH16 format.
Generally, depth output operates at a slower frame rate than standard color capture,
so the DEPTH16 and DEPTH_POINT_CLOUD formats will commonly have a stall duration that
should be accounted for (see {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration|ACAMERA_DEPTH_AVAILABLE_DEPTH_STALL_DURATIONS}).
On a device that supports both depth and color-based output, to enable smooth preview,
using a repeating burst is recommended, where a depth-output target is only included
once every N frames, where N is the ratio between preview output rate and depth output
rate, including depth stall time.
CONSTRAINED_HIGH_SPEED_VIDEO
The device supports constrained high speed video recording (frame rate >=120fps) use
case. The camera device will support high speed capture session created by {@link
android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}, which
only accepts high speed request lists created by {@link
android.hardware.camera2.CameraConstrainedHighSpeedCaptureSession#createHighSpeedRequestList}.
A camera device can still support high speed video streaming by advertising the high
speed FPS ranges in android.control.aeAvailableTargetFpsRanges. For this case, all
normal capture request per frame control and synchronization requirements will apply
to the high speed fps ranges, the same as all other fps ranges. This capability
describes the capability of a specialized operating mode with many limitations (see
below), which is only targeted at high speed video recording.
The supported high speed video sizes and fps ranges are specified in {@link
android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoFpsRanges}.
To get desired output frame rates, the application is only allowed to select video
size and FPS range combinations provided by {@link
android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoSizes}. The
fps range can be controlled via android.control.aeTargetFpsRange.
In this capability, the camera device will override aeMode, awbMode, and afMode to
ON, AUTO, and CONTINUOUS_VIDEO, respectively. All post-processing block mode
controls will be overridden to be FAST. Therefore, no manual control of capture
and post-processing parameters is possible. All other controls operate the
same as when android.control.mode == AUTO. This means that all other
android.control.* fields continue to work, such as
* android.control.aeTargetFpsRange
* android.control.aeExposureCompensation
* android.control.aeLock
* android.control.awbLock
* android.control.effectMode
* android.control.aeRegions
* android.control.afRegions
* android.control.awbRegions
* android.control.afTrigger
* android.control.aePrecaptureTrigger
Outside of android.control.*, the following controls will work:
* android.flash.mode (TORCH mode only, automatic flash for still capture will not
work since aeMode is ON)
* android.lens.opticalStabilizationMode (if it is supported)
* android.scaler.cropRegion
* android.statistics.faceDetectMode (if it is supported)
For high speed recording use case, the actual maximum supported frame rate may
be lower than what camera can output, depending on the destination Surfaces for
the image data. For example, if the destination surface is from video encoder,
the application need check if the video encoder is capable of supporting the
high frame rate for a given video size, or it will end up with lower recording
frame rate. If the destination surface is from preview window, the actual preview frame
rate will be bounded by the screen refresh rate.
The camera device will only support up to 2 high speed simultaneous output surfaces
(preview and recording surfaces) in this mode. Above controls will be effective only
if all of below conditions are true:
* The application creates a camera capture session with no more than 2 surfaces via
{@link
android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}. The
targeted surfaces must be preview surface (either from {@link
android.view.SurfaceView} or {@link android.graphics.SurfaceTexture}) or recording
surface(either from {@link android.media.MediaRecorder#getSurface} or {@link
android.media.MediaCodec#createInputSurface}).
* The stream sizes are selected from the sizes reported by
{@link android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoSizes}.
* The FPS ranges are selected from {@link
android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoFpsRanges}.
When above conditions are NOT satistied,
{@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}
will fail.
Switching to a FPS range that has different maximum FPS may trigger some camera device
reconfigurations, which may introduce extra latency. It is recommended that
the application avoids unnecessary maximum target FPS changes as much as possible
during high speed streaming.
MOTION_TRACKING
The camera device supports the MOTION_TRACKING value for
android.control.captureIntent, which limits maximum exposure time to 20 ms.
This limits the motion blur of capture images, resulting in better image tracking
results for use cases such as image stabilization or augmented reality.
LOGICAL_MULTI_CAMERA
The camera device is a logical camera backed by two or more physical cameras.
In API level 28, the physical cameras must also be exposed to the application via
{@link android.hardware.camera2.CameraManager#getCameraIdList}.
Starting from API level 29, some or all physical cameras may not be independently
exposed to the application, in which case the physical camera IDs will not be
available in {@link android.hardware.camera2.CameraManager#getCameraIdList}. But the
application can still query the physical cameras' characteristics by calling
{@link android.hardware.camera2.CameraManager#getCameraCharacteristics}. Additionally,
if a physical camera is hidden from camera ID list, the mandatory stream combinations
for that physical camera must be supported through the logical camera using physical
streams.
Combinations of logical and physical streams, or physical streams from different
physical cameras are not guaranteed. However, if the camera device supports
{@link CameraDevice#isSessionConfigurationSupported|ACameraDevice_isSessionConfigurationSupported},
application must be able to query whether a stream combination involving physical
streams is supported by calling
{@link CameraDevice#isSessionConfigurationSupported|ACameraDevice_isSessionConfigurationSupported}.
Camera application shouldn't assume that there are at most 1 rear camera and 1 front
camera in the system. For an application that switches between front and back cameras,
the recommendation is to switch between the first rear camera and the first front
camera in the list of supported camera devices.
This capability requires the camera device to support the following:
* The IDs of underlying physical cameras are returned via
{@link android.hardware.camera2.CameraCharacteristics#getPhysicalCameraIds}.
* This camera device must list static metadata
android.logicalMultiCamera.sensorSyncType in
{@link android.hardware.camera2.CameraCharacteristics}.
* The underlying physical cameras' static metadata must list the following entries,
so that the application can correlate pixels from the physical streams:
- android.lens.poseReference
- android.lens.poseRotation
- android.lens.poseTranslation
- android.lens.intrinsicCalibration
- android.lens.distortion
* The SENSOR_INFO_TIMESTAMP_SOURCE of the logical device and physical devices must be
the same.
* The logical camera must be LIMITED or higher device.
A logical camera device's dynamic metadata may contain
android.logicalMultiCamera.activePhysicalId to notify the application of the current
active physical camera Id. An active physical camera is the physical camera from which
the logical camera's main image data outputs (YUV or RAW) and metadata come from.
In addition, this serves as an indication which physical camera is used to output to
a RAW stream, or in case only physical cameras support RAW, which physical RAW stream
the application should request.
Logical camera's static metadata tags below describe the default active physical
camera. An active physical camera is default if it's used when application directly
uses requests built from a template. All templates will default to the same active
physical camera.
- android.sensor.info.sensitivityRange
- android.sensor.info.colorFilterArrangement
- android.sensor.info.exposureTimeRange
- android.sensor.info.maxFrameDuration
- android.sensor.info.physicalSize
- android.sensor.info.whiteLevel
- android.sensor.info.lensShadingApplied
- android.sensor.referenceIlluminant1
- android.sensor.referenceIlluminant2
- android.sensor.calibrationTransform1
- android.sensor.calibrationTransform2
- android.sensor.colorTransform1
- android.sensor.colorTransform2
- android.sensor.forwardMatrix1
- android.sensor.forwardMatrix2
- android.sensor.blackLevelPattern
- android.sensor.maxAnalogSensitivity
- android.sensor.opticalBlackRegions
- android.sensor.availableTestPatternModes
- android.lens.info.hyperfocalDistance
- android.lens.info.minimumFocusDistance
- android.lens.info.focusDistanceCalibration
- android.lens.poseRotation
- android.lens.poseTranslation
- android.lens.intrinsicCalibration
- android.lens.poseReference
- android.lens.distortion
The field of view of all non-RAW physical streams must be the same or as close as
possible to that of non-RAW logical streams. If the requested FOV is outside of the
range supported by the physical camera, the physical stream for that physical camera
will use either the maximum or minimum scaler crop region, depending on which one is
closer to the requested FOV. For example, for a logical camera with wide-tele lens
configuration where the wide lens is the default, if the logical camera's crop region
is set to maximum, the physical stream for the tele lens will be configured to its
maximum crop region. On the other hand, if the logical camera has a normal-wide lens
configuration where the normal lens is the default, when the logical camera's crop
region is set to maximum, the FOV of the logical streams will be that of the normal
lens. The FOV of the physical streams for the wide lens will be the same as the
logical stream, by making the crop region smaller than its active array size to
compensate for the smaller focal length.
Even if the underlying physical cameras have different RAW characteristics (such as
size or CFA pattern), a logical camera can still advertise RAW capability. In this
case, when the application configures a RAW stream, the camera device will make sure
the active physical camera will remain active to ensure consistent RAW output
behavior, and not switch to other physical cameras.
The capture request and result metadata tags required for backward compatible camera
functionalities will be solely based on the logical camera capabiltity. On the other
hand, the use of manual capture controls (sensor or post-processing) with a
logical camera may result in unexpected behavior when the HAL decides to switch
between physical cameras with different characteristics under the hood. For example,
when the application manually sets exposure time and sensitivity while zooming in,
the brightness of the camera images may suddenly change because HAL switches from one
physical camera to the other.
MONOCHROME
The camera device is a monochrome camera that doesn't contain a color filter array,
and for YUV_420_888 stream, the pixel values on U and V planes are all 128.
A MONOCHROME camera must support the guaranteed stream combinations required for
its device level and capabilities. Additionally, if the monochrome camera device
supports Y8 format, all mandatory stream combination requirements related to {@link
android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888 YUV_420_888} apply
to {@link android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8 Y8} as well. There are no
mandatory stream combination requirements with regard to
{@link android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8 Y8} for Bayer camera devices.
Starting from Android Q, the SENSOR_INFO_COLOR_FILTER_ARRANGEMENT of a MONOCHROME
camera will be either MONO or NIR.
SECURE_IMAGE_DATA
The camera device is capable of writing image data into a region of memory
inaccessible to Android userspace or the Android kernel, and only accessible to
trusted execution environments (TEE).
List of capabilities that this camera device
advertises as fully supporting.
A capability is a contract that the camera device makes in order
to be able to satisfy one or more use cases.
Listing a capability guarantees that the whole set of features
required to support a common use will all be available.
Using a subset of the functionality provided by an unsupported
capability may be possible on a specific camera device implementation;
to do this query each of android.request.availableRequestKeys,
android.request.availableResultKeys,
android.request.availableCharacteristicsKeys.
The following capabilities are guaranteed to be available on
android.info.supportedHardwareLevel `==` FULL devices:
* MANUAL_SENSOR
* MANUAL_POST_PROCESSING
Other capabilities may be available on either FULL or LIMITED
devices, but the application should query this key to be sure.
Additional constraint details per-capability will be available
in the Compatibility Test Suite.
Minimum baseline requirements required for the
BACKWARD_COMPATIBLE capability are not explicitly listed.
Instead refer to "BC" tags and the camera CTS tests in the
android.hardware.camera2.cts package.
Listed controls that can be either request or result (e.g.
android.sensor.exposureTime) must be available both in the
request and the result in order to be considered to be
capability-compliant.
For example, if the HAL claims to support MANUAL control,
then exposure time must be configurable via the request _and_
the actual exposure applied must be available via
the result.
If MANUAL_SENSOR is omitted, the HAL may choose to omit the
android.scaler.availableMinFrameDurations static property entirely.
For PRIVATE_REPROCESSING and YUV_REPROCESSING capabilities, see
hardware/libhardware/include/hardware/camera3.h Section 10 for more information.
Devices that support the MANUAL_SENSOR capability must support the
CAMERA3_TEMPLATE_MANUAL template defined in camera3.h.
Devices that support the PRIVATE_REPROCESSING capability or the
YUV_REPROCESSING capability must support the
CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template defined in camera3.h.
For DEPTH_OUTPUT, the depth-format keys
android.depth.availableDepthStreamConfigurations,
android.depth.availableDepthMinFrameDurations,
android.depth.availableDepthStallDurations must be available, in
addition to the other keys explicitly mentioned in the DEPTH_OUTPUT
enum notes. The entry android.depth.maxDepthSamples must be available
if the DEPTH_POINT_CLOUD format is supported (HAL pixel format BLOB, dataspace
DEPTH).
For a camera device with LOGICAL_MULTI_CAMERA capability, it should operate in the
same way as a physical camera device based on its hardware level and capabilities.
It's recommended that its feature set is superset of that of individual physical cameras.
* In camera1 API, to maintain application compatibility, for each camera facing, there
may be one or more {logical_camera_id, physical_camera_1_id, physical_camera_2_id, ...}
combinations, where logical_camera_id is composed of physical_camera_N_id, camera
framework will only advertise one camera id
(within the combinations for the particular facing) that is frontmost in the HAL
published camera id list.
For example, if HAL advertises 6 back facing camera IDs (ID0 to ID5), among which ID4
and ID5 are logical cameras backed by ID0+ID1 and ID2+ID3 respectively. In this case,
only ID0 will be available for camera1 API to use.
* Camera HAL is strongly recommended to advertise camera devices with best feature,
power, performance, and latency tradeoffs at the front of the camera id list.
* Camera HAL may switch between physical cameras depending on focalLength or cropRegion.
If physical cameras have different sizes, HAL must maintain a single logical camera
activeArraySize/pixelArraySize/preCorrectionActiveArraySize, and must do proper mapping
between logical camera and underlying physical cameras for all related metadata tags,
such as crop region, 3A regions, and intrinsicCalibration.
* Starting from HIDL ICameraDevice version 3.5, camera HAL must support
isStreamCombinationSupported for application to query whether a particular logical and
physical streams combination are supported.
A MONOCHROME camera device must also advertise BACKWARD_COMPATIBLE capability, and must
not advertise MANUAL_POST_PROCESSING capability.
* To maintain backward compatibility, the camera device must support all
BACKWARD_COMPATIBLE required keys. The android.control.awbAvailableModes key only contains
AUTO, and android.control.awbState are either CONVERGED or LOCKED depending on
android.control.awbLock.
* android.colorCorrection.mode, android.colorCorrection.transform, and
android.colorCorrection.gains must not be in available request and result keys.
As a result, the camera device cannot be a FULL device. However, the HAL can
still advertise other individual capabilites.
* If the device supports tonemap control, only android.tonemap.curveRed is used.
CurveGreen and curveBlue are no-ops.
In Android API level 28, a MONOCHROME camera device must not have RAW capability. From
API level 29, a camera is allowed to have both MONOCHROME and RAW capabilities.
n
A list of all keys that the camera device has available
to use with {@link android.hardware.camera2.CaptureRequest|ACaptureRequest}.
Attempting to set a key into a CaptureRequest that is not
listed here will result in an invalid request and will be rejected
by the camera device.
This field can be used to query the feature set of a camera device
at a more granular level than capabilities. This is especially
important for optional keys that are not listed under any capability
in android.request.availableCapabilities.
Vendor tags can be listed here. Vendor tag metadata should also
use the extensions C api (refer to camera3.h for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not consume any request tags that are not listed either
here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible
via
{@link android.hardware.camera2.CameraCharacteristics#getAvailableCaptureRequestKeys}.
n
A list of all keys that the camera device has available to use with {@link
android.hardware.camera2.CaptureResult|ACameraCaptureSession_captureCallback_result}.
Attempting to get a key from a CaptureResult that is not
listed here will always return a `null` value. Getting a key from
a CaptureResult that is listed here will generally never return a `null`
value.
The following keys may return `null` unless they are enabled:
* android.statistics.lensShadingMap (non-null iff android.statistics.lensShadingMapMode == ON)
(Those sometimes-null keys will nevertheless be listed here
if they are available.)
This field can be used to query the feature set of a camera device
at a more granular level than capabilities. This is especially
important for optional keys that are not listed under any capability
in android.request.availableCapabilities.
Tags listed here must always have an entry in the result metadata,
even if that size is 0 elements. Only array-type tags (e.g. lists,
matrices, strings) are allowed to have 0 elements.
Vendor tags can be listed here. Vendor tag metadata should also
use the extensions C api (refer to camera3.h for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not produce any result tags that are not listed either
here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible via {@link
android.hardware.camera2.CameraCharacteristics#getAvailableCaptureResultKeys}.
n
A list of all keys that the camera device has available to use with {@link
android.hardware.camera2.CameraCharacteristics|ACameraManager_getCameraCharacteristics}.
This entry follows the same rules as
android.request.availableResultKeys (except that it applies for
CameraCharacteristics instead of CaptureResult). See above for more
details.
Keys listed here must always have an entry in the static info metadata,
even if that size is 0 elements. Only array-type tags (e.g. lists,
matrices, strings) are allowed to have 0 elements.
Vendor tags can listed here. Vendor tag metadata should also use
the extensions C api (refer to camera3.h for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not have any tags in its static info that are not listed
either here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible
via {@link android.hardware.camera2.CameraCharacteristics#getKeys}.
n
A subset of the available request keys that the camera device
can pass as part of the capture session initialization.
This is a subset of android.request.availableRequestKeys which
contains a list of keys that are difficult to apply per-frame and
can result in unexpected delays when modified during the capture session
lifetime. Typical examples include parameters that require a
time-consuming hardware re-configuration or internal camera pipeline
change. For performance reasons we advise clients to pass their initial
values as part of
{@link SessionConfiguration#setSessionParameters|ACameraDevice_createCaptureSessionWithSessionParameters}.
Once the camera capture session is enabled it is also recommended to avoid
changing them from their initial values set in
{@link SessionConfiguration#setSessionParameters|ACameraDevice_createCaptureSessionWithSessionParameters}.
Control over session parameters can still be exerted in capture requests
but clients should be aware and expect delays during their application.
An example usage scenario could look like this:
* The camera client starts by quering the session parameter key list via
{@link android.hardware.camera2.CameraCharacteristics#getAvailableSessionKeys|ACameraManager_getCameraCharacteristics}.
* Before triggering the capture session create sequence, a capture request
must be built via
{@link CameraDevice#createCaptureRequest|ACameraDevice_createCaptureRequest}
using an appropriate template matching the particular use case.
* The client should go over the list of session parameters and check
whether some of the keys listed matches with the parameters that
they intend to modify as part of the first capture request.
* If there is no such match, the capture request can be passed
unmodified to
{@link SessionConfiguration#setSessionParameters|ACameraDevice_createCaptureSessionWithSessionParameters}.
* If matches do exist, the client should update the respective values
and pass the request to
{@link SessionConfiguration#setSessionParameters|ACameraDevice_createCaptureSessionWithSessionParameters}.
* After the capture session initialization completes the session parameter
key list can continue to serve as reference when posting or updating
further requests. As mentioned above further changes to session
parameters should ideally be avoided, if updates are necessary
however clients could expect a delay/glitch during the
parameter switch.
If android.control.aeTargetFpsRange is part of the session parameters and constrained high
speed mode is enabled, then only modifications of the maximum framerate value will be
monitored by the framework and can trigger camera re-configuration. For more information
about framerate ranges during constrained high speed sessions see
{@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}.
Vendor tags can be listed here. Vendor tag metadata should also
use the extensions C api (refer to
android.hardware.camera.device.V3_4.StreamConfiguration.sessionParams for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not consume any request tags in the session parameters that
are not listed either here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible
via
{@link android.hardware.camera2.CameraCharacteristics#getAvailableSessionKeys}.
n
A subset of the available request keys that can be overridden for
physical devices backing a logical multi-camera.
This is a subset of android.request.availableRequestKeys which contains a list
of keys that can be overridden using {@link CaptureRequest.Builder#setPhysicalCameraKey}.
The respective value of such request key can be obtained by calling
{@link CaptureRequest.Builder#getPhysicalCameraKey}. Capture requests that contain
individual physical device requests must be built via
{@link android.hardware.camera2.CameraDevice#createCaptureRequest(int, Set)}.
Vendor tags can be listed here. Vendor tag metadata should also
use the extensions C api (refer to
android.hardware.camera.device.V3_4.CaptureRequest.physicalCameraSettings for more
details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not consume any request tags in the session parameters that
are not listed either here or in the vendor tag list.
There should be no overlap between this set of keys and the available session keys
{@link android.hardware.camera2.CameraCharacteristics#getAvailableSessionKeys} along
with any other controls that can have impact on the dual-camera sync.
The public camera2 API will always make the vendor tags visible
via
{@link android.hardware.camera2.CameraCharacteristics#getAvailablePhysicalCameraRequestKeys}.
n
A list of camera characteristics keys that are only available
in case the camera client has camera permission.
The entry contains a subset of
{@link android.hardware.camera2.CameraCharacteristics#getKeys} that require camera clients
to acquire the {@link android.Manifest.permission#CAMERA} permission before calling
{@link android.hardware.camera2.CameraManager#getCameraCharacteristics}. If the
permission is not held by the camera client, then the values of the repsective properties
will not be present in {@link android.hardware.camera2.CameraCharacteristics}.
Do not set this property directly, camera service will overwrite any previous values.
4
The desired region of the sensor to read out for this capture.
Pixel coordinates relative to
android.sensor.info.activeArraySize or
android.sensor.info.preCorrectionActiveArraySize depending on distortion correction
capability and mode
This control can be used to implement digital zoom.
For devices not supporting android.distortionCorrection.mode control, the coordinate
system always follows that of android.sensor.info.activeArraySize, with `(0, 0)` being
the top-left pixel of the active array.
For devices supporting android.distortionCorrection.mode control, the coordinate
system depends on the mode being set.
When the distortion correction mode is OFF, the coordinate system follows
android.sensor.info.preCorrectionActiveArraySize, with
`(0, 0)` being the top-left pixel of the pre-correction active array.
When the distortion correction mode is not OFF, the coordinate system follows
android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array.
Output streams use this rectangle to produce their output,
cropping to a smaller region if necessary to maintain the
stream's aspect ratio, then scaling the sensor input to
match the output's configured resolution.
The crop region is applied after the RAW to other color
space (e.g. YUV) conversion. Since raw streams
(e.g. RAW16) don't have the conversion stage, they are not
croppable. The crop region will be ignored by raw streams.
For non-raw streams, any additional per-stream cropping will
be done to maximize the final pixel area of the stream.
For example, if the crop region is set to a 4:3 aspect
ratio, then 4:3 streams will use the exact crop
region. 16:9 streams will further crop vertically
(letterbox).
Conversely, if the crop region is set to a 16:9, then 4:3
outputs will crop horizontally (pillarbox), and 16:9
streams will match exactly. These additional crops will
be centered within the crop region.
If the coordinate system is android.sensor.info.activeArraySize, the width and height
of the crop region cannot be set to be smaller than
`floor( activeArraySize.width / android.scaler.availableMaxDigitalZoom )` and
`floor( activeArraySize.height / android.scaler.availableMaxDigitalZoom )`, respectively.
If the coordinate system is android.sensor.info.preCorrectionActiveArraySize, the width
and height of the crop region cannot be set to be smaller than
`floor( preCorrectionActiveArraySize.width / android.scaler.availableMaxDigitalZoom )`
and
`floor( preCorrectionActiveArraySize.height / android.scaler.availableMaxDigitalZoom )`,
respectively.
The camera device may adjust the crop region to account
for rounding and other hardware requirements; the final
crop region used will be included in the output capture
result.
The data representation is int[4], which maps to (left, top, width, height).
The output streams must maintain square pixels at all
times, no matter what the relative aspect ratios of the
crop region and the stream are. Negative values for
corner are allowed for raw output if full pixel array is
larger than active pixel array. Width and height may be
rounded to nearest larger supportable width, especially
for raw output, where only a few fixed scales may be
possible.
For a set of output streams configured, if the sensor output is cropped to a smaller
size than pre-correction active array size, the HAL need follow below cropping rules:
* The HAL need handle the cropRegion as if the sensor crop size is the effective
pre-correction active array size. More specifically, the HAL must transform the request
cropRegion from android.sensor.info.preCorrectionActiveArraySize to the sensor cropped
pixel area size in this way:
1. Translate the requested cropRegion w.r.t., the left top corner of the sensor
cropped pixel area by (tx, ty),
where `ty = sensorCrop.top * (sensorCrop.height / preCorrectionActiveArraySize.height)`
and `tx = sensorCrop.left * (sensorCrop.width / preCorrectionActiveArraySize.width)`.
The (sensorCrop.top, sensorCrop.left) is the coordinate based off the
android.sensor.info.activeArraySize.
2. Scale the width and height of requested cropRegion with scaling factor of
sensorCrop.width/preCorrectionActiveArraySize.width and sensorCrop.height/preCorrectionActiveArraySize.height
respectively.
Once this new cropRegion is calculated, the HAL must use this region to crop the image
with regard to the sensor crop size (effective pre-correction active array size). The
HAL still need follow the general cropping rule for this new cropRegion and effective
pre-correction active array size.
* The HAL must report the cropRegion with regard to android.sensor.info.preCorrectionActiveArraySize.
The HAL need convert the new cropRegion generated above w.r.t., full pre-correction
active array size. The reported cropRegion may be slightly different with the requested
cropRegion since the HAL may adjust the crop region to account for rounding, conversion
error, or other hardware limitations.
HAL2.x uses only (x, y, width)
n
RAW16
RAW16 is a standard, cross-platform format for raw image
buffers with 16-bit pixels.
Buffers of this format are typically expected to have a
Color Filter Array (CFA) layout, which is given in
android.sensor.info.colorFilterArrangement. Sensors with
CFAs that are not representable by a format in
android.sensor.info.colorFilterArrangement should not
use this format.
Buffers of this format will also follow the constraints given for
RAW_OPAQUE buffers, but with relaxed performance constraints.
This format is intended to give users access to the full contents
of the buffers coming directly from the image sensor prior to any
cropping or scaling operations, and all coordinate systems for
metadata used for this format are relative to the size of the
active region of the image sensor before any geometric distortion
correction has been applied (i.e.
android.sensor.info.preCorrectionActiveArraySize). Supported
dimensions for this format are limited to the full dimensions of
the sensor (e.g. either android.sensor.info.pixelArraySize or
android.sensor.info.preCorrectionActiveArraySize will be the
only supported output size).
See android.scaler.availableInputOutputFormatsMap for
the full set of performance guarantees.
RAW_OPAQUE
RAW_OPAQUE (or
{@link android.graphics.ImageFormat#RAW_PRIVATE RAW_PRIVATE}
as referred in public API) is a format for raw image buffers
coming from an image sensor.
The actual structure of buffers of this format is
platform-specific, but must follow several constraints:
1. No image post-processing operations may have been applied to
buffers of this type. These buffers contain raw image data coming
directly from the image sensor.
1. If a buffer of this format is passed to the camera device for
reprocessing, the resulting images will be identical to the images
produced if the buffer had come directly from the sensor and was
processed with the same settings.
The intended use for this format is to allow access to the native
raw format buffers coming directly from the camera sensor without
any additional conversions or decrease in framerate.
See android.scaler.availableInputOutputFormatsMap for the full set of
performance guarantees.
YV12
YCrCb 4:2:0 Planar
YCrCb_420_SP
NV21
IMPLEMENTATION_DEFINED
System internal format, not application-accessible
YCbCr_420_888
Flexible YUV420 Format
BLOB
JPEG format
RAW10
RAW10
RAW12
RAW12
Y8
Y8
The list of image formats that are supported by this
camera device for output streams.
Not used in HALv3 or newer
All camera devices will support JPEG and YUV_420_888 formats.
When set to YUV_420_888, application can access the YUV420 data directly.
These format values are from HAL_PIXEL_FORMAT_* in
system/core/include/system/graphics.h.
When IMPLEMENTATION_DEFINED is used, the platform
gralloc module will select a format based on the usage flags provided
by the camera HAL device and the other endpoint of the stream. It is
usually used by preview and recording streams, where the application doesn't
need access the image data.
YCbCr_420_888 format must be supported by the HAL. When an image stream
needs CPU/application direct access, this format will be used. For a MONOCHROME
camera device, the pixel value of Cb and Cr planes is 128.
The BLOB format must be supported by the HAL. This is used for the JPEG stream.
A RAW_OPAQUE buffer should contain only pixel data. It is strongly
recommended that any information used by the camera device when
processing images is fully expressed by the result metadata
for that image buffer.
n
The minimum frame duration that is supported
for each resolution in android.scaler.availableJpegSizes.
Not used in HALv3 or newer
Nanoseconds
TODO: Remove property.
This corresponds to the minimum steady-state frame duration when only
that JPEG stream is active and captured in a burst, with all
processing (typically in android.*.mode) set to FAST.
When multiple streams are configured, the minimum
frame duration will be >= max(individual stream min
durations)
n
2
The JPEG resolutions that are supported by this camera device.
Not used in HALv3 or newer
TODO: Remove property.
The resolutions are listed as `(width, height)` pairs. All camera devices will support
sensor maximum resolution (defined by android.sensor.info.activeArraySize).
The HAL must include sensor maximum resolution
(defined by android.sensor.info.activeArraySize),
and should include half/quarter of sensor maximum resolution.
The maximum ratio between both active area width
and crop region width, and active area height and
crop region height, for android.scaler.cropRegion.
Zoom scale factor
>=1
This represents the maximum amount of zooming possible by
the camera device, or equivalently, the minimum cropping
window size.
Crop regions that have a width or height that is smaller
than this ratio allows will be rounded up to the minimum
allowed size by the camera device.
n
For each available processed output size (defined in
android.scaler.availableProcessedSizes), this property lists the
minimum supportable frame duration for that size.
Not used in HALv3 or newer
Nanoseconds
This should correspond to the frame duration when only that processed
stream is active, with all processing (typically in android.*.mode)
set to FAST.
When multiple streams are configured, the minimum frame duration will
be >= max(individual stream min durations).
n
2
The resolutions available for use with
processed output streams, such as YV12, NV12, and
platform opaque YUV/RGB streams to the GPU or video
encoders.
Not used in HALv3 or newer
The resolutions are listed as `(width, height)` pairs.
For a given use case, the actual maximum supported resolution
may be lower than what is listed here, depending on the destination
Surface for the image data. For example, for recording video,
the video encoder chosen may have a maximum size limit (e.g. 1080p)
smaller than what the camera (e.g. maximum resolution is 3264x2448)
can provide.
Please reference the documentation for the image data destination to
check if it limits the maximum size for image data.
For FULL capability devices (`android.info.supportedHardwareLevel == FULL`),
the HAL must include all JPEG sizes listed in android.scaler.availableJpegSizes
and each below resolution if it is smaller than or equal to the sensor
maximum resolution (if they are not listed in JPEG sizes already):
* 240p (320 x 240)
* 480p (640 x 480)
* 720p (1280 x 720)
* 1080p (1920 x 1080)
For LIMITED capability devices (`android.info.supportedHardwareLevel == LIMITED`),
the HAL only has to list up to the maximum video size supported by the devices.
n
For each available raw output size (defined in
android.scaler.availableRawSizes), this property lists the minimum
supportable frame duration for that size.
Not used in HALv3 or newer
Nanoseconds
Should correspond to the frame duration when only the raw stream is
active.
When multiple streams are configured, the minimum
frame duration will be >= max(individual stream min
durations)
n
2
The resolutions available for use with raw
sensor output streams, listed as width,
height
Not used in HALv3 or newer
The mapping of image formats that are supported by this
camera device for input streams, to their corresponding output formats.
All camera devices with at least 1
android.request.maxNumInputStreams will have at least one
available input format.
The camera device will support the following map of formats,
if its dependent capability (android.request.availableCapabilities) is supported:
Input Format | Output Format | Capability
:-------------------------------------------------|:--------------------------------------------------|:----------
{@link android.graphics.ImageFormat#PRIVATE} | {@link android.graphics.ImageFormat#JPEG} | PRIVATE_REPROCESSING
{@link android.graphics.ImageFormat#PRIVATE} | {@link android.graphics.ImageFormat#YUV_420_888} | PRIVATE_REPROCESSING
{@link android.graphics.ImageFormat#YUV_420_888} | {@link android.graphics.ImageFormat#JPEG} | YUV_REPROCESSING
{@link android.graphics.ImageFormat#YUV_420_888} | {@link android.graphics.ImageFormat#YUV_420_888} | YUV_REPROCESSING
PRIVATE refers to a device-internal format that is not directly application-visible. A
PRIVATE input surface can be acquired by {@link android.media.ImageReader#newInstance}
with {@link android.graphics.ImageFormat#PRIVATE} as the format.
For a PRIVATE_REPROCESSING-capable camera device, using the PRIVATE format as either input
or output will never hurt maximum frame rate (i.e. {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration
getOutputStallDuration(ImageFormat.PRIVATE, size)} is always 0),
Attempting to configure an input stream with output streams not
listed as available in this map is not valid.
Additionally, if the camera device is MONOCHROME with Y8 support, it will also support
the following map of formats if its dependent capability
(android.request.availableCapabilities) is supported:
Input Format | Output Format | Capability
:-------------------------------------------------|:--------------------------------------------------|:----------
{@link android.graphics.ImageFormat#PRIVATE} | {@link android.graphics.ImageFormat#Y8} | PRIVATE_REPROCESSING
{@link android.graphics.ImageFormat#Y8} | {@link android.graphics.ImageFormat#JPEG} | YUV_REPROCESSING
{@link android.graphics.ImageFormat#Y8} | {@link android.graphics.ImageFormat#Y8} | YUV_REPROCESSING
For the formats, see `system/core/include/system/graphics.h` for a definition
of the image format enumerations. The PRIVATE format refers to the
HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED format. The HAL could determine
the actual format by using the gralloc usage flags.
For ZSL use case in particular, the HAL could choose appropriate format (partially
processed YUV or RAW based format) by checking the format and GRALLOC_USAGE_HW_CAMERA_ZSL.
See camera3.h for more details.
This value is encoded as a variable-size array-of-arrays.
The inner array always contains `[format, length, ...]` where
`...` has `length` elements. An inner array is followed by another
inner array if the total metadata entry size hasn't yet been exceeded.
A code sample to read/write this encoding (with a device that
supports reprocessing IMPLEMENTATION_DEFINED to YUV_420_888, and JPEG,
and reprocessing YUV_420_888 to YUV_420_888 and JPEG):
// reading
int32_t* contents = &entry.i32[0];
for (size_t i = 0; i < entry.count; ) {
int32_t format = contents[i++];
int32_t length = contents[i++];
int32_t output_formats[length];
memcpy(&output_formats[0], &contents[i],
length * sizeof(int32_t));
i += length;
}
// writing (static example, PRIVATE_REPROCESSING + YUV_REPROCESSING)
int32_t[] contents = {
IMPLEMENTATION_DEFINED, 2, YUV_420_888, BLOB,
YUV_420_888, 2, YUV_420_888, BLOB,
};
update_camera_metadata_entry(metadata, index, &contents[0],
sizeof(contents)/sizeof(contents[0]), &updated_entry);
If the HAL claims to support any of the capabilities listed in the
above details, then it must also support all the input-output
combinations listed for that capability. It can optionally support
additional formats if it so chooses.
n
4
OUTPUT
INPUT
The available stream configurations that this
camera device supports
(i.e. format, width, height, output/input stream).
The configurations are listed as `(format, width, height, input?)`
tuples.
For a given use case, the actual maximum supported resolution
may be lower than what is listed here, depending on the destination
Surface for the image data. For example, for recording video,
the video encoder chosen may have a maximum size limit (e.g. 1080p)
smaller than what the camera (e.g. maximum resolution is 3264x2448)
can provide.
Please reference the documentation for the image data destination to
check if it limits the maximum size for image data.
Not all output formats may be supported in a configuration with
an input stream of a particular format. For more details, see
android.scaler.availableInputOutputFormatsMap.
The following table describes the minimum required output stream
configurations based on the hardware level
(android.info.supportedHardwareLevel):
Format | Size | Hardware Level | Notes
:-------------:|:--------------------------------------------:|:--------------:|:--------------:
JPEG | android.sensor.info.activeArraySize | Any |
JPEG | 1920x1080 (1080p) | Any | if 1080p <= activeArraySize
JPEG | 1280x720 (720) | Any | if 720p <= activeArraySize
JPEG | 640x480 (480p) | Any | if 480p <= activeArraySize
JPEG | 320x240 (240p) | Any | if 240p <= activeArraySize
YUV_420_888 | all output sizes available for JPEG | FULL |
YUV_420_888 | all output sizes available for JPEG, up to the maximum video size | LIMITED |
IMPLEMENTATION_DEFINED | same as YUV_420_888 | Any |
Refer to android.request.availableCapabilities for additional
mandatory stream configurations on a per-capability basis.
Exception on 176x144 (QCIF) resolution: camera devices usually have a fixed capability for
downscaling from larger resolution to smaller, and the QCIF resolution sometimes is not
fully supported due to this limitation on devices with high-resolution image sensors.
Therefore, trying to configure a QCIF resolution stream together with any other
stream larger than 1920x1080 resolution (either width or height) might not be supported,
and capture session creation will fail if it is not.
It is recommended (but not mandatory) to also include half/quarter
of sensor maximum resolution for JPEG formats (regardless of hardware
level).
(The following is a rewording of the above required table):
For JPEG format, the sizes may be restricted by below conditions:
* The HAL may choose the aspect ratio of each Jpeg size to be one of well known ones
(e.g. 4:3, 16:9, 3:2 etc.). If the sensor maximum resolution
(defined by android.sensor.info.activeArraySize) has an aspect ratio other than these,
it does not have to be included in the supported JPEG sizes.
* Some hardware JPEG encoders may have pixel boundary alignment requirements, such as
the dimensions being a multiple of 16.
Therefore, the maximum JPEG size may be smaller than sensor maximum resolution.
However, the largest JPEG size must be as close as possible to the sensor maximum
resolution given above constraints. It is required that after aspect ratio adjustments,
additional size reduction due to other issues must be less than 3% in area. For example,
if the sensor maximum resolution is 3280x2464, if the maximum JPEG size has aspect
ratio 4:3, the JPEG encoder alignment requirement is 16, the maximum JPEG size will be
3264x2448.
For FULL capability devices (`android.info.supportedHardwareLevel == FULL`),
the HAL must include all YUV_420_888 sizes that have JPEG sizes listed
here as output streams.
It must also include each below resolution if it is smaller than or
equal to the sensor maximum resolution (for both YUV_420_888 and JPEG
formats), as output streams:
* 240p (320 x 240)
* 480p (640 x 480)
* 720p (1280 x 720)
* 1080p (1920 x 1080)
For LIMITED capability devices
(`android.info.supportedHardwareLevel == LIMITED`),
the HAL only has to list up to the maximum video size
supported by the device.
Regardless of hardware level, every output resolution available for
YUV_420_888 must also be available for IMPLEMENTATION_DEFINED.
This supercedes the following fields, which are now deprecated:
* availableFormats
* available[Processed,Raw,Jpeg]Sizes
4
n
This lists the minimum frame duration for each
format/size combination.
(format, width, height, ns) x n
This should correspond to the frame duration when only that
stream is active, with all processing (typically in android.*.mode)
set to either OFF or FAST.
When multiple streams are used in a request, the minimum frame
duration will be max(individual stream min durations).
The minimum frame duration of a stream (of a particular format, size)
is the same regardless of whether the stream is input or output.
See android.sensor.frameDuration and
android.scaler.availableStallDurations for more details about
calculating the max frame rate.
4
n
This lists the maximum stall duration for each
output format/size combination.
(format, width, height, ns) x n
A stall duration is how much extra time would get added
to the normal minimum frame duration for a repeating request
that has streams with non-zero stall.
For example, consider JPEG captures which have the following
characteristics:
* JPEG streams act like processed YUV streams in requests for which
they are not included; in requests in which they are directly
referenced, they act as JPEG streams. This is because supporting a
JPEG stream requires the underlying YUV data to always be ready for
use by a JPEG encoder, but the encoder will only be used (and impact
frame duration) on requests that actually reference a JPEG stream.
* The JPEG processor can run concurrently to the rest of the camera
pipeline, but cannot process more than 1 capture at a time.
In other words, using a repeating YUV request would result
in a steady frame rate (let's say it's 30 FPS). If a single
JPEG request is submitted periodically, the frame rate will stay
at 30 FPS (as long as we wait for the previous JPEG to return each
time). If we try to submit a repeating YUV + JPEG request, then
the frame rate will drop from 30 FPS.
In general, submitting a new request with a non-0 stall time
stream will _not_ cause a frame rate drop unless there are still
outstanding buffers for that stream from previous requests.
Submitting a repeating request with streams (call this `S`)
is the same as setting the minimum frame duration from
the normal minimum frame duration corresponding to `S`, added with
the maximum stall duration for `S`.
If interleaving requests with and without a stall duration,
a request will stall by the maximum of the remaining times
for each can-stall stream with outstanding buffers.
This means that a stalling request will not have an exposure start
until the stall has completed.
This should correspond to the stall duration when only that stream is
active, with all processing (typically in android.*.mode) set to FAST
or OFF. Setting any of the processing modes to HIGH_QUALITY
effectively results in an indeterminate stall duration for all
streams in a request (the regular stall calculation rules are
ignored).
The following formats may always have a stall duration:
* {@link android.graphics.ImageFormat#JPEG|AIMAGE_FORMAT_JPEG}
* {@link android.graphics.ImageFormat#RAW_SENSOR|AIMAGE_FORMAT_RAW16}
The following formats will never have a stall duration:
* {@link android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888}
* {@link android.graphics.ImageFormat#RAW10|AIMAGE_FORMAT_RAW10}
* {@link android.graphics.ImageFormat#RAW12|AIMAGE_FORMAT_RAW12}
* {@link android.graphics.ImageFormat#Y8|AIMAGE_FORMAT_Y8}
All other formats may or may not have an allowed stall duration on
a per-capability basis; refer to android.request.availableCapabilities
for more details.
See android.sensor.frameDuration for more information about
calculating the max frame rate (absent stalls).
If possible, it is recommended that all non-JPEG formats
(such as RAW16) should not have a stall duration. RAW10, RAW12, RAW_OPAQUE
and IMPLEMENTATION_DEFINED must not have stall durations.
The available stream configurations that this
camera device supports; also includes the minimum frame durations
and the stall durations for each format/size combination.
All camera devices will support sensor maximum resolution (defined by
android.sensor.info.activeArraySize) for the JPEG format.
For a given use case, the actual maximum supported resolution
may be lower than what is listed here, depending on the destination
Surface for the image data. For example, for recording video,
the video encoder chosen may have a maximum size limit (e.g. 1080p)
smaller than what the camera (e.g. maximum resolution is 3264x2448)
can provide.
Please reference the documentation for the image data destination to
check if it limits the maximum size for image data.
The following table describes the minimum required output stream
configurations based on the hardware level
(android.info.supportedHardwareLevel):
Format | Size | Hardware Level | Notes
:-------------------------------------------------:|:--------------------------------------------:|:--------------:|:--------------:
{@link android.graphics.ImageFormat#JPEG} | android.sensor.info.activeArraySize (*1) | Any |
{@link android.graphics.ImageFormat#JPEG} | 1920x1080 (1080p) | Any | if 1080p <= activeArraySize
{@link android.graphics.ImageFormat#JPEG} | 1280x720 (720p) | Any | if 720p <= activeArraySize
{@link android.graphics.ImageFormat#JPEG} | 640x480 (480p) | Any | if 480p <= activeArraySize
{@link android.graphics.ImageFormat#JPEG} | 320x240 (240p) | Any | if 240p <= activeArraySize
{@link android.graphics.ImageFormat#YUV_420_888} | all output sizes available for JPEG | FULL |
{@link android.graphics.ImageFormat#YUV_420_888} | all output sizes available for JPEG, up to the maximum video size | LIMITED |
{@link android.graphics.ImageFormat#PRIVATE} | same as YUV_420_888 | Any |
Refer to android.request.availableCapabilities and {@link
android.hardware.camera2.CameraDevice#createCaptureSession} for additional mandatory
stream configurations on a per-capability basis.
*1: For JPEG format, the sizes may be restricted by below conditions:
* The HAL may choose the aspect ratio of each Jpeg size to be one of well known ones
(e.g. 4:3, 16:9, 3:2 etc.). If the sensor maximum resolution
(defined by android.sensor.info.activeArraySize) has an aspect ratio other than these,
it does not have to be included in the supported JPEG sizes.
* Some hardware JPEG encoders may have pixel boundary alignment requirements, such as
the dimensions being a multiple of 16.
Therefore, the maximum JPEG size may be smaller than sensor maximum resolution.
However, the largest JPEG size will be as close as possible to the sensor maximum
resolution given above constraints. It is required that after aspect ratio adjustments,
additional size reduction due to other issues must be less than 3% in area. For example,
if the sensor maximum resolution is 3280x2464, if the maximum JPEG size has aspect
ratio 4:3, and the JPEG encoder alignment requirement is 16, the maximum JPEG size will be
3264x2448.
Exception on 176x144 (QCIF) resolution: camera devices usually have a fixed capability on
downscaling from larger resolution to smaller ones, and the QCIF resolution can sometimes
not be fully supported due to this limitation on devices with high-resolution image
sensors. Therefore, trying to configure a QCIF resolution stream together with any other
stream larger than 1920x1080 resolution (either width or height) might not be supported,
and capture session creation will fail if it is not.
Do not set this property directly
(it is synthetic and will not be available at the HAL layer);
set the android.scaler.availableStreamConfigurations instead.
Not all output formats may be supported in a configuration with
an input stream of a particular format. For more details, see
android.scaler.availableInputOutputFormatsMap.
It is recommended (but not mandatory) to also include half/quarter
of sensor maximum resolution for JPEG formats (regardless of hardware
level).
(The following is a rewording of the above required table):
The HAL must include sensor maximum resolution (defined by
android.sensor.info.activeArraySize).
For FULL capability devices (`android.info.supportedHardwareLevel == FULL`),
the HAL must include all YUV_420_888 sizes that have JPEG sizes listed
here as output streams.
It must also include each below resolution if it is smaller than or
equal to the sensor maximum resolution (for both YUV_420_888 and JPEG
formats), as output streams:
* 240p (320 x 240)
* 480p (640 x 480)
* 720p (1280 x 720)
* 1080p (1920 x 1080)
For LIMITED capability devices
(`android.info.supportedHardwareLevel == LIMITED`),
the HAL only has to list up to the maximum video size
supported by the device.
Regardless of hardware level, every output resolution available for
YUV_420_888 must also be available for IMPLEMENTATION_DEFINED.
This supercedes the following fields, which are now deprecated:
* availableFormats
* available[Processed,Raw,Jpeg]Sizes
CENTER_ONLY
The camera device only supports centered crop regions.
FREEFORM
The camera device supports arbitrarily chosen crop regions.
The crop type that this camera device supports.
When passing a non-centered crop region (android.scaler.cropRegion) to a camera
device that only supports CENTER_ONLY cropping, the camera device will move the
crop region to the center of the sensor active array (android.sensor.info.activeArraySize)
and keep the crop region width and height unchanged. The camera device will return the
final used crop region in metadata result android.scaler.cropRegion.
Camera devices that support FREEFORM cropping will support any crop region that
is inside of the active array. The camera device will apply the same crop region and
return the final used crop region in capture result metadata android.scaler.cropRegion.
LEGACY capability devices will only support CENTER_ONLY cropping.
n
5
PREVIEW
Preview must only include non-stalling processed stream configurations with
output formats like
{@link android.graphics.ImageFormat#YUV_420_888|AIMAGE_FORMAT_YUV_420_888},
{@link android.graphics.ImageFormat#PRIVATE|AIMAGE_FORMAT_PRIVATE}, etc.
RECORD
Video record must include stream configurations that match the advertised
supported media profiles {@link android.media.CamcorderProfile} with
IMPLEMENTATION_DEFINED format.
VIDEO_SNAPSHOT
Video snapshot must include stream configurations at least as big as
the maximum RECORD resolutions and only with
{@link android.graphics.ImageFormat#JPEG|AIMAGE_FORMAT_JPEG JPEG output format}.
Additionally the configurations shouldn't cause preview glitches and also be able to
run at 30 fps.
SNAPSHOT
Recommended snapshot stream configurations must include at least one with
size close to android.sensor.info.activeArraySize and
{@link android.graphics.ImageFormat#JPEG|AIMAGE_FORMAT_JPEG JPEG output format}.
Taking into account restrictions on aspect ratio, alignment etc. the area of the
maximum suggested size shouldn’t be less than 97% of the sensor array size area.
ZSL
If supported, recommended input stream configurations must only be advertised with
ZSL along with other processed and/or stalling output formats.
RAW
If supported, recommended raw stream configurations must only include RAW based
output formats.
LOW_LATENCY_SNAPSHOT
If supported, the recommended low latency stream configurations must have
end-to-end latency that does not exceed 200 ms. under standard operating conditions
(reasonable light levels, not loaded system) and using template
TEMPLATE_STILL_CAPTURE. This is primarily for listing configurations for the
{@link android.graphics.ImageFormat#JPEG|AIMAGE_FORMAT_JPEG JPEG output format}
however other supported output formats can be added as well.
PUBLIC_END
VENDOR_START
Vendor defined use cases. These depend on the vendor implementation.
Recommended stream configurations for common client use cases.
Optional subset of the android.scaler.availableStreamConfigurations that contains
similar tuples listed as
(i.e. width, height, format, output/input stream, usecase bit field).
Camera devices will be able to suggest particular stream configurations which are
power and performance efficient for specific use cases. For more information about
retrieving the suggestions see
{@link android.hardware.camera2.CameraCharacteristics#getRecommendedStreamConfigurationMap}.
The data representation is int[5], which maps to
(width, height, format, output/input stream, usecase bit field). The array can be
parsed using the following pseudo code:
struct StreamConfiguration {
int32_t format;
int32_t width;
int32_t height;
int32_t isInput; };
void getPreferredStreamConfigurations(
int32_t *array, size_t count, int32_t usecaseId,
Vector < StreamConfiguration > * scs) {
const size_t STREAM_CONFIGURATION_SIZE = 5;
const size_t STREAM_WIDTH_OFFSET = 0;
const size_t STREAM_HEIGHT_OFFSET = 1;
const size_t STREAM_FORMAT_OFFSET = 2;
const size_t STREAM_IS_INPUT_OFFSET = 3;
const size_t STREAM_USECASE_BITMAP_OFFSET = 4;
for (size_t i = 0; i < count; i+= STREAM_CONFIGURATION_SIZE) {
int32_t width = array[i + STREAM_WIDTH_OFFSET];
int32_t height = array[i + STREAM_HEIGHT_OFFSET];
int32_t format = array[i + STREAM_FORMAT_OFFSET];
int32_t isInput = array[i + STREAM_IS_INPUT_OFFSET];
int32_t supportedUsecases = array[i + STREAM_USECASE_BITMAP_OFFSET];
if (supportedUsecases & (1 << usecaseId)) {
StreamConfiguration sc = {format, width, height, isInput};
scs->add(sc);
}
}
}
There are some requirements that need to be considered regarding the usecases and the
suggested configurations:
* If android.scaler.availableRecommendedStreamConfigurations is set, then recommended
stream configurations must be present for all mandatory usecases PREVIEW,
SNAPSHOT, RECORD, VIDEO_SNAPSHOT. ZSL and RAW are
required depending on device capabilities see android.request.availableCapabilities.
* Non-existing usecases and non-vendor usecases within the range
(RAW : VENDOR_START] are prohibited as well as stream configurations not
present in the exhaustive android.scaler.availableStreamConfigurations list.
For example, in case the camera device supports only 4K and 1080p and both resolutions are
recommended for the mandatory usecases except preview which can run efficiently only
on 1080p. The array may look like this:
[3840, 2160, HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED,
ANDROID_SCALER_AVAILABLE_STREAM_CONFIGURATIONS_OUTPUT,
(1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_RECORD |
1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_SNAPSHOT |
1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_VIDEO_SNAPSHOT),
1920, 1080, HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED,
ANDROID_SCALER_AVAILABLE_STREAM_CONFIGURATIONS_OUTPUT,
(1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_PREVIEW |
1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_RECORD |
1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_SNAPSHOT |
1<< ANDROID_SCALER_AVAILABLE_RECOMMENDED_STREAM_CONFIGURATIONS_VIDEO_SNAPSHOT)]
Recommended mappings of image formats that are supported by this
camera device for input streams, to their corresponding output formats.
This is a recommended subset of the complete list of mappings found in
android.scaler.availableInputOutputFormatsMap. The same requirements apply here as well.
The list however doesn't need to contain all available and supported mappings. Instead of
this developers must list only recommended and efficient entries.
If set, the information will be available in the ZERO_SHUTTER_LAG recommended stream
configuration see
{@link android.hardware.camera2.CameraCharacteristics#getRecommendedStreamConfigurationMap}.
For a code sample of the required data encoding please check
android.scaler.availableInputOutputFormatsMap.
n
An array of mandatory stream combinations generated according to the camera device
{@link android.hardware.camera2.CameraCharacteristics#INFO_SUPPORTED_HARDWARE_LEVEL}
and {@link android.hardware.camera2.CameraCharacteristics#REQUEST_AVAILABLE_CAPABILITIES}.
This is an app-readable conversion of the mandatory stream combination
{@link android.hardware.camera2.CameraDevice#createCaptureSession tables}.
The array of
{@link android.hardware.camera2.params.MandatoryStreamCombination combinations} is
generated according to the documented
{@link android.hardware.camera2.CameraDevice#createCaptureSession guideline} based on
specific device level and capabilities.
Clients can use the array as a quick reference to find an appropriate camera stream
combination.
As per documentation, the stream combinations with given PREVIEW, RECORD and
MAXIMUM resolutions and anything smaller from the list given by
{@link android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes} are
guaranteed to work.
For a physical camera not independently exposed in
{@link android.hardware.camera2.CameraManager#getCameraIdList}, the mandatory stream
combinations for that physical camera Id are also generated, so that the application can
configure them as physical streams via the logical camera.
The mandatory stream combination array will be {@code null} in case the device is not
backward compatible.
Do not set this property directly
(it is synthetic and will not be available at the HAL layer).
Duration each pixel is exposed to
light.
Nanoseconds
android.sensor.info.exposureTimeRange
If the sensor can't expose this exact duration, it will shorten the
duration exposed to the nearest possible value (rather than expose longer).
The final exposure time used will be available in the output capture result.
This control is only effective if android.control.aeMode or android.control.mode is set to
OFF; otherwise the auto-exposure algorithm will override this value.
Duration from start of frame exposure to
start of next frame exposure.
Nanoseconds
See android.sensor.info.maxFrameDuration, {@link
android.hardware.camera2.params.StreamConfigurationMap|ACAMERA_SCALER_AVAILABLE_MIN_FRAME_DURATIONS}.
The duration is capped to `max(duration, exposureTime + overhead)`.
The maximum frame rate that can be supported by a camera subsystem is
a function of many factors:
* Requested resolutions of output image streams
* Availability of binning / skipping modes on the imager
* The bandwidth of the imager interface
* The bandwidth of the various ISP processing blocks
Since these factors can vary greatly between different ISPs and
sensors, the camera abstraction tries to represent the bandwidth
restrictions with as simple a model as possible.
The model presented has the following characteristics:
* The image sensor is always configured to output the smallest
resolution possible given the application's requested output stream
sizes. The smallest resolution is defined as being at least as large
as the largest requested output stream size; the camera pipeline must
never digitally upsample sensor data when the crop region covers the
whole sensor. In general, this means that if only small output stream
resolutions are configured, the sensor can provide a higher frame
rate.
* Since any request may use any or all the currently configured
output streams, the sensor and ISP must be configured to support
scaling a single capture to all the streams at the same time. This
means the camera pipeline must be ready to produce the largest
requested output size without any delay. Therefore, the overall
frame rate of a given configured stream set is governed only by the
largest requested stream resolution.
* Using more than one output stream in a request does not affect the
frame duration.
* Certain format-streams may need to do additional background processing
before data is consumed/produced by that stream. These processors
can run concurrently to the rest of the camera pipeline, but
cannot process more than 1 capture at a time.
The necessary information for the application, given the model above, is provided via
{@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration|ACAMERA_SCALER_AVAILABLE_MIN_FRAME_DURATIONS}.
These are used to determine the maximum frame rate / minimum frame duration that is
possible for a given stream configuration.
Specifically, the application can use the following rules to
determine the minimum frame duration it can request from the camera
device:
1. Let the set of currently configured input/output streams be called `S`.
1. Find the minimum frame durations for each stream in `S`, by looking it up in {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration|ACAMERA_SCALER_AVAILABLE_MIN_FRAME_DURATIONS}
(with its respective size/format). Let this set of frame durations be called `F`.
1. For any given request `R`, the minimum frame duration allowed for `R` is the maximum
out of all values in `F`. Let the streams used in `R` be called `S_r`.
If none of the streams in `S_r` have a stall time (listed in {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration|ACAMERA_SCALER_AVAILABLE_STALL_DURATIONS}
using its respective size/format), then the frame duration in `F` determines the steady
state frame rate that the application will get if it uses `R` as a repeating request. Let
this special kind of request be called `Rsimple`.
A repeating request `Rsimple` can be _occasionally_ interleaved by a single capture of a
new request `Rstall` (which has at least one in-use stream with a non-0 stall time) and if
`Rstall` has the same minimum frame duration this will not cause a frame rate loss if all
buffers from the previous `Rstall` have already been delivered.
For more details about stalling, see {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration|ACAMERA_SCALER_AVAILABLE_STALL_DURATIONS}.
This control is only effective if android.control.aeMode or android.control.mode is set to
OFF; otherwise the auto-exposure algorithm will override this value.
For more details about stalling, see
android.scaler.availableStallDurations.
The amount of gain applied to sensor data
before processing.
ISO arithmetic units
android.sensor.info.sensitivityRange
The sensitivity is the standard ISO sensitivity value,
as defined in ISO 12232:2006.
The sensitivity must be within android.sensor.info.sensitivityRange, and
if if it less than android.sensor.maxAnalogSensitivity, the camera device
is guaranteed to use only analog amplification for applying the gain.
If the camera device cannot apply the exact sensitivity
requested, it will reduce the gain to the nearest supported
value. The final sensitivity used will be available in the
output capture result.
This control is only effective if android.control.aeMode or android.control.mode is set to
OFF; otherwise the auto-exposure algorithm will override this value.
ISO 12232:2006 REI method is acceptable.
4
The area of the image sensor which corresponds to active pixels after any geometric
distortion correction has been applied.
Pixel coordinates on the image sensor
This is the rectangle representing the size of the active region of the sensor (i.e.
the region that actually receives light from the scene) after any geometric correction
has been applied, and should be treated as the maximum size in pixels of any of the
image output formats aside from the raw formats.
This rectangle is defined relative to the full pixel array; (0,0) is the top-left of
the full pixel array, and the size of the full pixel array is given by
android.sensor.info.pixelArraySize.
The coordinate system for most other keys that list pixel coordinates, including
android.scaler.cropRegion, is defined relative to the active array rectangle given in
this field, with `(0, 0)` being the top-left of this rectangle.
The active array may be smaller than the full pixel array, since the full array may
include black calibration pixels or other inactive regions.
For devices that do not support android.distortionCorrection.mode control, the active
array must be the same as android.sensor.info.preCorrectionActiveArraySize.
For devices that support android.distortionCorrection.mode control, the active array must
be enclosed by android.sensor.info.preCorrectionActiveArraySize. The difference between
pre-correction active array and active array accounts for scaling or cropping caused
by lens geometric distortion correction.
In general, application should always refer to active array size for controls like
metering regions or crop region. Two exceptions are when the application is dealing with
RAW image buffers (RAW_SENSOR, RAW10, RAW12 etc), or when application explicitly set
android.distortionCorrection.mode to OFF. In these cases, application should refer
to android.sensor.info.preCorrectionActiveArraySize.
The data representation is `int[4]`, which maps to `(left, top, width, height)`.
This array contains `(xmin, ymin, width, height)`. The `(xmin, ymin)` must be
>= `(0,0)`.
The `(width, height)` must be <= `android.sensor.info.pixelArraySize`.
2
Range of sensitivities for android.sensor.sensitivity supported by this
camera device.
Min <= 100, Max >= 800
The values are the standard ISO sensitivity values,
as defined in ISO 12232:2006.
RGGB
GRBG
GBRG
BGGR
RGB
Sensor is not Bayer; output has 3 16-bit
values for each pixel, instead of just 1 16-bit value
per pixel.
MONO
Sensor doesn't have any Bayer color filter.
Such sensor captures visible light in monochrome. The exact weighting and
wavelengths captured is not specified, but generally only includes the visible
frequencies. This value implies a MONOCHROME camera.
NIR
Sensor has a near infrared filter capturing light with wavelength between
roughly 750nm and 1400nm, and the same filter covers the whole sensor array. This
value implies a MONOCHROME camera.
The arrangement of color filters on sensor;
represents the colors in the top-left 2x2 section of
the sensor, in reading order, for a Bayer camera, or the
light spectrum it captures for MONOCHROME camera.
Starting from Android Q, the colorFilterArrangement for a MONOCHROME camera must be
single color patterns, such as MONO or NIR.
2
The range of image exposure times for android.sensor.exposureTime supported
by this camera device.
Nanoseconds
The minimum exposure time will be less than 100 us. For FULL
capability devices (android.info.supportedHardwareLevel == FULL),
the maximum exposure time will be greater than 100ms.
For FULL capability devices (android.info.supportedHardwareLevel == FULL),
The maximum of the range SHOULD be at least 1 second (1e9), MUST be at least
100ms.
The maximum possible frame duration (minimum frame rate) for
android.sensor.frameDuration that is supported this camera device.
Nanoseconds
For FULL capability devices
(android.info.supportedHardwareLevel == FULL), at least 100ms.
Attempting to use frame durations beyond the maximum will result in the frame
duration being clipped to the maximum. See that control for a full definition of frame
durations.
Refer to {@link
android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration|ACAMERA_SCALER_AVAILABLE_MIN_FRAME_DURATIONS}
for the minimum frame duration values.
For FULL capability devices (android.info.supportedHardwareLevel == FULL),
The maximum of the range SHOULD be at least
1 second (1e9), MUST be at least 100ms (100e6).
android.sensor.info.maxFrameDuration must be greater or
equal to the android.sensor.info.exposureTimeRange max
value (since exposure time overrides frame duration).
Available minimum frame durations for JPEG must be no greater
than that of the YUV_420_888/IMPLEMENTATION_DEFINED
minimum frame durations (for that respective size).
Since JPEG processing is considered offline and can take longer than
a single uncompressed capture, refer to
android.scaler.availableStallDurations
for details about encoding this scenario.
2
The physical dimensions of the full pixel
array.
Millimeters
This is the physical size of the sensor pixel
array defined by android.sensor.info.pixelArraySize.
Needed for FOV calculation for old API
2
Dimensions of the full pixel array, possibly
including black calibration pixels.
Pixels
The pixel count of the full pixel array of the image sensor, which covers
android.sensor.info.physicalSize area. This represents the full pixel dimensions of
the raw buffers produced by this sensor.
If a camera device supports raw sensor formats, either this or
android.sensor.info.preCorrectionActiveArraySize is the maximum dimensions for the raw
output formats listed in {@link
android.hardware.camera2.params.StreamConfigurationMap|ACAMERA_SCALER_AVAILABLE_STREAM_CONFIGURATIONS}
(this depends on whether or not the image sensor returns buffers containing pixels that
are not part of the active array region for blacklevel calibration or other purposes).
Some parts of the full pixel array may not receive light from the scene,
or be otherwise inactive. The android.sensor.info.preCorrectionActiveArraySize key
defines the rectangle of active pixels that will be included in processed image
formats.
Maximum raw value output by sensor.
> 255 (8-bit output)
This specifies the fully-saturated encoding level for the raw
sample values from the sensor. This is typically caused by the
sensor becoming highly non-linear or clipping. The minimum for
each channel is specified by the offset in the
android.sensor.blackLevelPattern key.
The white level is typically determined either by sensor bit depth
(8-14 bits is expected), or by the point where the sensor response
becomes too non-linear to be useful. The default value for this is
maximum representable value for a 16-bit raw sample (2^16 - 1).
The white level values of captured images may vary for different
capture settings (e.g., android.sensor.sensitivity). This key
represents a coarse approximation for such case. It is recommended
to use android.sensor.dynamicWhiteLevel for captures when supported
by the camera device, which provides more accurate white level values.
The full bit depth of the sensor must be available in the raw data,
so the value for linear sensors should not be significantly lower
than maximum raw value supported, i.e. 2^(sensor bits per pixel).
UNKNOWN
Timestamps from android.sensor.timestamp are in nanoseconds and monotonic,
but can not be compared to timestamps from other subsystems
(e.g. accelerometer, gyro etc.), or other instances of the same or different
camera devices in the same system. Timestamps between streams and results for
a single camera instance are comparable, and the timestamps for all buffers
and the result metadata generated by a single capture are identical.
REALTIME
Timestamps from android.sensor.timestamp are in the same timebase as
{@link android.os.SystemClock#elapsedRealtimeNanos},
and they can be compared to other timestamps using that base.
The time base source for sensor capture start timestamps.
The timestamps provided for captures are always in nanoseconds and monotonic, but
may not based on a time source that can be compared to other system time sources.
This characteristic defines the source for the timestamps, and therefore whether they
can be compared against other system time sources/timestamps.
For camera devices implement UNKNOWN, the camera framework expects that the timestamp
source to be SYSTEM_TIME_MONOTONIC. For camera devices implement REALTIME, the camera
framework expects that the timestamp source to be SYSTEM_TIME_BOOTTIME. See
system/core/include/utils/Timers.h for the definition of SYSTEM_TIME_MONOTONIC and
SYSTEM_TIME_BOOTTIME. Note that HAL must follow above expectation; otherwise video
recording might suffer unexpected behavior.
Also, camera devices which implement REALTIME must pass the ITS sensor fusion test which
tests the alignment between camera timestamps and gyro sensor timestamps.
FALSE
TRUE
Whether the RAW images output from this camera device are subject to
lens shading correction.
If TRUE, all images produced by the camera device in the RAW image formats will
have lens shading correction already applied to it. If FALSE, the images will
not be adjusted for lens shading correction.
See android.request.maxNumOutputRaw for a list of RAW image formats.
This key will be `null` for all devices do not report this information.
Devices with RAW capability will always report this information in this key.
4
The area of the image sensor which corresponds to active pixels prior to the
application of any geometric distortion correction.
Pixel coordinates on the image sensor
This is the rectangle representing the size of the active region of the sensor (i.e.
the region that actually receives light from the scene) before any geometric correction
has been applied, and should be treated as the active region rectangle for any of the
raw formats. All metadata associated with raw processing (e.g. the lens shading
correction map, and radial distortion fields) treats the top, left of this rectangle as
the origin, (0,0).
The size of this region determines the maximum field of view and the maximum number of
pixels that an image from this sensor can contain, prior to the application of
geometric distortion correction. The effective maximum pixel dimensions of a
post-distortion-corrected image is given by the android.sensor.info.activeArraySize
field, and the effective maximum field of view for a post-distortion-corrected image
can be calculated by applying the geometric distortion correction fields to this
rectangle, and cropping to the rectangle given in android.sensor.info.activeArraySize.
E.g. to calculate position of a pixel, (x,y), in a processed YUV output image with the
dimensions in android.sensor.info.activeArraySize given the position of a pixel,
(x', y'), in the raw pixel array with dimensions give in
android.sensor.info.pixelArraySize:
1. Choose a pixel (x', y') within the active array region of the raw buffer given in
android.sensor.info.preCorrectionActiveArraySize, otherwise this pixel is considered
to be outside of the FOV, and will not be shown in the processed output image.
1. Apply geometric distortion correction to get the post-distortion pixel coordinate,
(x_i, y_i). When applying geometric correction metadata, note that metadata for raw
buffers is defined relative to the top, left of the
android.sensor.info.preCorrectionActiveArraySize rectangle.
1. If the resulting corrected pixel coordinate is within the region given in
android.sensor.info.activeArraySize, then the position of this pixel in the
processed output image buffer is `(x_i - activeArray.left, y_i - activeArray.top)`,
when the top, left coordinate of that buffer is treated as (0, 0).
Thus, for pixel x',y' = (25, 25) on a sensor where android.sensor.info.pixelArraySize
is (100,100), android.sensor.info.preCorrectionActiveArraySize is (10, 10, 100, 100),
android.sensor.info.activeArraySize is (20, 20, 80, 80), and the geometric distortion
correction doesn't change the pixel coordinate, the resulting pixel selected in
pixel coordinates would be x,y = (25, 25) relative to the top,left of the raw buffer
with dimensions given in android.sensor.info.pixelArraySize, and would be (5, 5)
relative to the top,left of post-processed YUV output buffer with dimensions given in
android.sensor.info.activeArraySize.
The currently supported fields that correct for geometric distortion are:
1. android.lens.distortion.
If the camera device doesn't support geometric distortion correction, or all of the
geometric distortion fields are no-ops, this rectangle will be the same as the
post-distortion-corrected rectangle given in android.sensor.info.activeArraySize.
This rectangle is defined relative to the full pixel array; (0,0) is the top-left of
the full pixel array, and the size of the full pixel array is given by
android.sensor.info.pixelArraySize.
The pre-correction active array may be smaller than the full pixel array, since the
full array may include black calibration pixels or other inactive regions.
The data representation is `int[4]`, which maps to `(left, top, width, height)`.
This array contains `(xmin, ymin, width, height)`. The `(xmin, ymin)` must be
>= `(0,0)`.
The `(width, height)` must be <= `android.sensor.info.pixelArraySize`.
If omitted by the HAL implementation, the camera framework will assume that this is
the same as the post-correction active array region given in
android.sensor.info.activeArraySize.
DAYLIGHT
FLUORESCENT
TUNGSTEN
Incandescent light
FLASH
FINE_WEATHER
CLOUDY_WEATHER
SHADE
DAYLIGHT_FLUORESCENT
D 5700 - 7100K
DAY_WHITE_FLUORESCENT
N 4600 - 5400K
COOL_WHITE_FLUORESCENT
W 3900 - 4500K
WHITE_FLUORESCENT
WW 3200 - 3700K
STANDARD_A
STANDARD_B
STANDARD_C
D55
D65
D75
D50
ISO_STUDIO_TUNGSTEN
The standard reference illuminant used as the scene light source when
calculating the android.sensor.colorTransform1,
android.sensor.calibrationTransform1, and
android.sensor.forwardMatrix1 matrices.
The values in this key correspond to the values defined for the
EXIF LightSource tag. These illuminants are standard light sources
that are often used calibrating camera devices.
If this key is present, then android.sensor.colorTransform1,
android.sensor.calibrationTransform1, and
android.sensor.forwardMatrix1 will also be present.
Some devices may choose to provide a second set of calibration
information for improved quality, including
android.sensor.referenceIlluminant2 and its corresponding matrices.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
The first reference illuminant (android.sensor.referenceIlluminant1)
and corresponding matrices must be present to support the RAW capability
and DNG output.
When producing raw images with a color profile that has only been
calibrated against a single light source, it is valid to omit
android.sensor.referenceIlluminant2 along with the
android.sensor.colorTransform2, android.sensor.calibrationTransform2,
and android.sensor.forwardMatrix2 matrices.
If only android.sensor.referenceIlluminant1 is included, it should be
chosen so that it is representative of typical scene lighting. In
general, D50 or DAYLIGHT will be chosen for this case.
If both android.sensor.referenceIlluminant1 and
android.sensor.referenceIlluminant2 are included, they should be
chosen to represent the typical range of scene lighting conditions.
In general, low color temperature illuminant such as Standard-A will
be chosen for the first reference illuminant and a higher color
temperature illuminant such as D65 will be chosen for the second
reference illuminant.
The standard reference illuminant used as the scene light source when
calculating the android.sensor.colorTransform2,
android.sensor.calibrationTransform2, and
android.sensor.forwardMatrix2 matrices.
Any value listed in android.sensor.referenceIlluminant1
See android.sensor.referenceIlluminant1 for more details.
If this key is present, then android.sensor.colorTransform2,
android.sensor.calibrationTransform2, and
android.sensor.forwardMatrix2 will also be present.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
3
3
A per-device calibration transform matrix that maps from the
reference sensor colorspace to the actual device sensor colorspace.
This matrix is used to correct for per-device variations in the
sensor colorspace, and is used for processing raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a per-device calibration transform that maps colors
from reference sensor color space (i.e. the "golden module"
colorspace) into this camera device's native sensor color
space under the first reference illuminant
(android.sensor.referenceIlluminant1).
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
3
3
A per-device calibration transform matrix that maps from the
reference sensor colorspace to the actual device sensor colorspace
(this is the colorspace of the raw buffer data).
This matrix is used to correct for per-device variations in the
sensor colorspace, and is used for processing raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a per-device calibration transform that maps colors
from reference sensor color space (i.e. the "golden module"
colorspace) into this camera device's native sensor color
space under the second reference illuminant
(android.sensor.referenceIlluminant2).
This matrix will only be present if the second reference
illuminant is present.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
3
3
A matrix that transforms color values from CIE XYZ color space to
reference sensor color space.
This matrix is used to convert from the standard CIE XYZ color
space to the reference sensor colorspace, and is used when processing
raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a color transform matrix that maps colors from the CIE
XYZ color space to the reference sensor color space (i.e. the
"golden module" colorspace) under the first reference illuminant
(android.sensor.referenceIlluminant1).
The white points chosen in both the reference sensor color space
and the CIE XYZ colorspace when calculating this transform will
match the standard white point for the first reference illuminant
(i.e. no chromatic adaptation will be applied by this transform).
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
3
3
A matrix that transforms color values from CIE XYZ color space to
reference sensor color space.
This matrix is used to convert from the standard CIE XYZ color
space to the reference sensor colorspace, and is used when processing
raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a color transform matrix that maps colors from the CIE
XYZ color space to the reference sensor color space (i.e. the
"golden module" colorspace) under the second reference illuminant
(android.sensor.referenceIlluminant2).
The white points chosen in both the reference sensor color space
and the CIE XYZ colorspace when calculating this transform will
match the standard white point for the second reference illuminant
(i.e. no chromatic adaptation will be applied by this transform).
This matrix will only be present if the second reference
illuminant is present.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
3
3
A matrix that transforms white balanced camera colors from the reference
sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.
This matrix is used to convert to the standard CIE XYZ colorspace, and
is used when processing raw buffer data.
This matrix is expressed as a 3x3 matrix in row-major-order, and contains
a color transform matrix that maps white balanced colors from the
reference sensor color space to the CIE XYZ color space with a D50 white
point.
Under the first reference illuminant (android.sensor.referenceIlluminant1)
this matrix is chosen so that the standard white point for this reference
illuminant in the reference sensor colorspace is mapped to D50 in the
CIE XYZ colorspace.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
3
3
A matrix that transforms white balanced camera colors from the reference
sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.
This matrix is used to convert to the standard CIE XYZ colorspace, and
is used when processing raw buffer data.
This matrix is expressed as a 3x3 matrix in row-major-order, and contains
a color transform matrix that maps white balanced colors from the
reference sensor color space to the CIE XYZ color space with a D50 white
point.
Under the second reference illuminant (android.sensor.referenceIlluminant2)
this matrix is chosen so that the standard white point for this reference
illuminant in the reference sensor colorspace is mapped to D50 in the
CIE XYZ colorspace.
This matrix will only be present if the second reference
illuminant is present.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
Gain factor from electrons to raw units when
ISO=100
4
A fixed black level offset for each of the color filter arrangement
(CFA) mosaic channels.
>= 0 for each.
This key specifies the zero light value for each of the CFA mosaic
channels in the camera sensor. The maximal value output by the
sensor is represented by the value in android.sensor.info.whiteLevel.
The values are given in the same order as channels listed for the CFA
layout key (see android.sensor.info.colorFilterArrangement), i.e. the
nth value given corresponds to the black level offset for the nth
color channel listed in the CFA.
The black level values of captured images may vary for different
capture settings (e.g., android.sensor.sensitivity). This key
represents a coarse approximation for such case. It is recommended to
use android.sensor.dynamicBlackLevel or use pixels from
android.sensor.opticalBlackRegions directly for captures when
supported by the camera device, which provides more accurate black
level values. For raw capture in particular, it is recommended to use
pixels from android.sensor.opticalBlackRegions to calculate black
level values for each frame.
For a MONOCHROME camera device, all of the 2x2 channels must have the same values.
The values are given in row-column scan order, with the first value
corresponding to the element of the CFA in row=0, column=0.
Maximum sensitivity that is implemented
purely through analog gain.
For android.sensor.sensitivity values less than or
equal to this, all applied gain must be analog. For
values above this, the gain applied can be a mix of analog and
digital.
Clockwise angle through which the output image needs to be rotated to be
upright on the device screen in its native orientation.
Degrees of clockwise rotation; always a multiple of
90
0, 90, 180, 270
Also defines the direction of rolling shutter readout, which is from top to bottom in
the sensor's coordinate system.
3
The number of input samples for each dimension of
android.sensor.profileHueSatMap.
Hue >= 1,
Saturation >= 2,
Value >= 1
The number of input samples for the hue, saturation, and value
dimension of android.sensor.profileHueSatMap. The order of the
dimensions given is hue, saturation, value; where hue is the 0th
element.
Time at start of exposure of first
row of the image sensor active array, in nanoseconds.
Nanoseconds
> 0
The timestamps are also included in all image
buffers produced for the same capture, and will be identical
on all the outputs.
When android.sensor.info.timestampSource `==` UNKNOWN,
the timestamps measure time since an unspecified starting point,
and are monotonically increasing. They can be compared with the
timestamps for other captures from the same camera device, but are
not guaranteed to be comparable to any other time source.
When android.sensor.info.timestampSource `==` REALTIME, the
timestamps measure time in the same timebase as {@link
android.os.SystemClock#elapsedRealtimeNanos}, and they can
be compared to other timestamps from other subsystems that
are using that base.
For reprocessing, the timestamp will match the start of exposure of
the input image, i.e. {@link CaptureResult#SENSOR_TIMESTAMP the
timestamp} in the TotalCaptureResult that was used to create the
reprocess capture request.
All timestamps must be in reference to the kernel's
CLOCK_BOOTTIME monotonic clock, which properly accounts for
time spent asleep. This allows for synchronization with
sensors that continue to operate while the system is
otherwise asleep.
If android.sensor.info.timestampSource `==` REALTIME,
The timestamp must be synchronized with the timestamps from other
sensor subsystems that are using the same timebase.
For reprocessing, the input image's start of exposure can be looked up
with android.sensor.timestamp from the metadata included in the
capture request.
The temperature of the sensor, sampled at the time
exposure began for this frame.
The thermal diode being queried should be inside the sensor PCB, or
somewhere close to it.
Celsius
Optional. This value is missing if no temperature is available.
3
The estimated camera neutral color in the native sensor colorspace at
the time of capture.
This value gives the neutral color point encoded as an RGB value in the
native sensor color space. The neutral color point indicates the
currently estimated white point of the scene illumination. It can be
used to interpolate between the provided color transforms when
processing raw sensor data.
The order of the values is R, G, B; where R is in the lowest index.
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
2
CFA Channels
Noise model coefficients for each CFA mosaic channel.
This key contains two noise model coefficients for each CFA channel
corresponding to the sensor amplification (S) and sensor readout
noise (O). These are given as pairs of coefficients for each channel
in the same order as channels listed for the CFA layout key
(see android.sensor.info.colorFilterArrangement). This is
represented as an array of Pair<Double, Double>, where
the first member of the Pair at index n is the S coefficient and the
second member is the O coefficient for the nth color channel in the CFA.
These coefficients are used in a two parameter noise model to describe
the amount of noise present in the image for each CFA channel. The
noise model used here is:
N(x) = sqrt(Sx + O)
Where x represents the recorded signal of a CFA channel normalized to
the range [0, 1], and S and O are the noise model coeffiecients for
that channel.
A more detailed description of the noise model can be found in the
Adobe DNG specification for the NoiseProfile tag.
For a MONOCHROME camera, there is only one color channel. So the noise model coefficients
will only contain one S and one O.
For a CFA layout of RGGB, the list of coefficients would be given as
an array of doubles S0,O0,S1,O1,..., where S0 and O0 are the coefficients
for the red channel, S1 and O1 are the coefficients for the first green
channel, etc.
hue_samples
saturation_samples
value_samples
3
A mapping containing a hue shift, saturation scale, and value scale
for each pixel.
The hue shift is given in degrees; saturation and value scale factors are
unitless and are between 0 and 1 inclusive
hue_samples, saturation_samples, and value_samples are given in
android.sensor.profileHueSatMapDimensions.
Each entry of this map contains three floats corresponding to the
hue shift, saturation scale, and value scale, respectively; where the
hue shift has the lowest index. The map entries are stored in the key
in nested loop order, with the value divisions in the outer loop, the
hue divisions in the middle loop, and the saturation divisions in the
inner loop. All zero input saturation entries are required to have a
value scale factor of 1.0.
samples
2
A list of x,y samples defining a tone-mapping curve for gamma adjustment.
Each sample has an input range of `[0, 1]` and an output range of
`[0, 1]`. The first sample is required to be `(0, 0)`, and the last
sample is required to be `(1, 1)`.
This key contains a default tone curve that can be applied while
processing the image as a starting point for user adjustments.
The curve is specified as a list of value pairs in linear gamma.
The curve is interpolated using a cubic spline.
The worst-case divergence between Bayer green channels.
>= 0
This value is an estimate of the worst case split between the
Bayer green channels in the red and blue rows in the sensor color
filter array.
The green split is calculated as follows:
1. A 5x5 pixel (or larger) window W within the active sensor array is
chosen. The term 'pixel' here is taken to mean a group of 4 Bayer
mosaic channels (R, Gr, Gb, B). The location and size of the window
chosen is implementation defined, and should be chosen to provide a
green split estimate that is both representative of the entire image
for this camera sensor, and can be calculated quickly.
1. The arithmetic mean of the green channels from the red
rows (mean_Gr) within W is computed.
1. The arithmetic mean of the green channels from the blue
rows (mean_Gb) within W is computed.
1. The maximum ratio R of the two means is computed as follows:
`R = max((mean_Gr + 1)/(mean_Gb + 1), (mean_Gb + 1)/(mean_Gr + 1))`
The ratio R is the green split divergence reported for this property,
which represents how much the green channels differ in the mosaic
pattern. This value is typically used to determine the treatment of
the green mosaic channels when demosaicing.
The green split value can be roughly interpreted as follows:
* R < 1.03 is a negligible split (<3% divergence).
* 1.20 <= R >= 1.03 will require some software
correction to avoid demosaic errors (3-20% divergence).
* R > 1.20 will require strong software correction to produce
a usuable image (>20% divergence).
Starting from Android Q, this key will not be present for a MONOCHROME camera, even if
the camera device has RAW capability.
The green split given may be a static value based on prior
characterization of the camera sensor using the green split
calculation method given here over a large, representative, sample
set of images. Other methods of calculation that produce equivalent
results, and can be interpreted in the same manner, may be used.
4
A pixel `[R, G_even, G_odd, B]` that supplies the test pattern
when android.sensor.testPatternMode is SOLID_COLOR.
Each color channel is treated as an unsigned 32-bit integer.
The camera device then uses the most significant X bits
that correspond to how many bits are in its Bayer raw sensor
output.
For example, a sensor with RAW10 Bayer output would use the
10 most significant bits from each color channel.
OFF
No test pattern mode is used, and the camera
device returns captures from the image sensor.
This is the default if the key is not set.
SOLID_COLOR
Each pixel in `[R, G_even, G_odd, B]` is replaced by its
respective color channel provided in
android.sensor.testPatternData.
For example:
android.testPatternData = [0, 0xFFFFFFFF, 0xFFFFFFFF, 0]
All green pixels are 100% green. All red/blue pixels are black.
android.testPatternData = [0xFFFFFFFF, 0, 0xFFFFFFFF, 0]
All red pixels are 100% red. Only the odd green pixels
are 100% green. All blue pixels are 100% black.
COLOR_BARS
All pixel data is replaced with an 8-bar color pattern.
The vertical bars (left-to-right) are as follows:
* 100% white
* yellow
* cyan
* green
* magenta
* red
* blue
* black
In general the image would look like the following:
W Y C G M R B K
W Y C G M R B K
W Y C G M R B K
W Y C G M R B K
W Y C G M R B K
. . . . . . . .
. . . . . . . .
. . . . . . . .
(B = Blue, K = Black)
Each bar should take up 1/8 of the sensor pixel array width.
When this is not possible, the bar size should be rounded
down to the nearest integer and the pattern can repeat
on the right side.
Each bar's height must always take up the full sensor
pixel array height.
Each pixel in this test pattern must be set to either
0% intensity or 100% intensity.
COLOR_BARS_FADE_TO_GRAY
The test pattern is similar to COLOR_BARS, except that
each bar should start at its specified color at the top,
and fade to gray at the bottom.
Furthermore each bar is further subdivided into a left and
right half. The left half should have a smooth gradient,
and the right half should have a quantized gradient.
In particular, the right half's should consist of blocks of the
same color for 1/16th active sensor pixel array width.
The least significant bits in the quantized gradient should
be copied from the most significant bits of the smooth gradient.
The height of each bar should always be a multiple of 128.
When this is not the case, the pattern should repeat at the bottom
of the image.
PN9
All pixel data is replaced by a pseudo-random sequence
generated from a PN9 512-bit sequence (typically implemented
in hardware with a linear feedback shift register).
The generator should be reset at the beginning of each frame,
and thus each subsequent raw frame with this test pattern should
be exactly the same as the last.
CUSTOM1
The first custom test pattern. All custom patterns that are
available only on this camera device are at least this numeric
value.
All of the custom test patterns will be static
(that is the raw image must not vary from frame to frame).
When enabled, the sensor sends a test pattern instead of
doing a real exposure from the camera.
android.sensor.availableTestPatternModes
When a test pattern is enabled, all manual sensor controls specified
by android.sensor.* will be ignored. All other controls should
work as normal.
For example, if manual flash is enabled, flash firing should still
occur (and that the test pattern remain unmodified, since the flash
would not actually affect it).
Defaults to OFF.
All test patterns are specified in the Bayer domain.
The HAL may choose to substitute test patterns from the sensor
with test patterns from on-device memory. In that case, it should be
indistinguishable to the ISP whether the data came from the
sensor interconnect bus (such as CSI2) or memory.
n
List of sensor test pattern modes for android.sensor.testPatternMode
supported by this camera device.
Any value listed in android.sensor.testPatternMode
Defaults to OFF, and always includes OFF if defined.
All custom modes must be >= CUSTOM1.
Duration between the start of first row exposure
and the start of last row exposure.
Nanoseconds
>= 0 and <
{@link android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration}.
This is the exposure time skew between the first and last
row exposure start times. The first row and the last row are
the first and last rows inside of the
android.sensor.info.activeArraySize.
For typical camera sensors that use rolling shutters, this is also equivalent
to the frame readout time.
The HAL must report `0` if the sensor is using global shutter, where all pixels begin
exposure at the same time.
4
num_regions
List of disjoint rectangles indicating the sensor
optically shielded black pixel regions.
In most camera sensors, the active array is surrounded by some
optically shielded pixel areas. By blocking light, these pixels
provides a reliable black reference for black level compensation
in active array region.
This key provides a list of disjoint rectangles specifying the
regions of optically shielded (with metal shield) black pixel
regions if the camera device is capable of reading out these black
pixels in the output raw images. In comparison to the fixed black
level values reported by android.sensor.blackLevelPattern, this key
may provide a more accurate way for the application to calculate
black level of each captured raw images.
When this key is reported, the android.sensor.dynamicBlackLevel and
android.sensor.dynamicWhiteLevel will also be reported.
The data representation is `int[4]`, which maps to `(left, top, width, height)`.
This array contains (xmin, ymin, width, height). The (xmin, ymin)
must be >= (0,0) and <=
android.sensor.info.pixelArraySize. The (width, height) must be
<= android.sensor.info.pixelArraySize. Each region must be
outside the region reported by
android.sensor.info.preCorrectionActiveArraySize.
The HAL must report minimal number of disjoint regions for the
optically shielded back pixel regions. For example, if a region can
be covered by one rectangle, the HAL must not split this region into
multiple rectangles.
4
A per-frame dynamic black level offset for each of the color filter
arrangement (CFA) mosaic channels.
>= 0 for each.
Camera sensor black levels may vary dramatically for different
capture settings (e.g. android.sensor.sensitivity). The fixed black
level reported by android.sensor.blackLevelPattern may be too
inaccurate to represent the actual value on a per-frame basis. The
camera device internal pipeline relies on reliable black level values
to process the raw images appropriately. To get the best image
quality, the camera device may choose to estimate the per frame black
level values either based on optically shielded black regions
(android.sensor.opticalBlackRegions) or its internal model.
This key reports the camera device estimated per-frame zero light
value for each of the CFA mosaic channels in the camera sensor. The
android.sensor.blackLevelPattern may only represent a coarse
approximation of the actual black level values. This value is the
black level used in camera device internal image processing pipeline
and generally more accurate than the fixed black level values.
However, since they are estimated values by the camera device, they
may not be as accurate as the black level values calculated from the
optical black pixels reported by android.sensor.opticalBlackRegions.
The values are given in the same order as channels listed for the CFA
layout key (see android.sensor.info.colorFilterArrangement), i.e. the
nth value given corresponds to the black level offset for the nth
color channel listed in the CFA.
For a MONOCHROME camera, all of the 2x2 channels must have the same values.
This key will be available if android.sensor.opticalBlackRegions is available or the
camera device advertises this key via {@link
android.hardware.camera2.CameraCharacteristics#getAvailableCaptureResultKeys|ACAMERA_REQUEST_AVAILABLE_RESULT_KEYS}.
The values are given in row-column scan order, with the first value
corresponding to the element of the CFA in row=0, column=0.
Maximum raw value output by sensor for this frame.
>= 0
Since the android.sensor.blackLevelPattern may change for different
capture settings (e.g., android.sensor.sensitivity), the white
level will change accordingly. This key is similar to
android.sensor.info.whiteLevel, but specifies the camera device
estimated white level for each frame.
This key will be available if android.sensor.opticalBlackRegions is
available or the camera device advertises this key via
{@link android.hardware.camera2.CameraCharacteristics#getAvailableCaptureRequestKeys|ACAMERA_REQUEST_AVAILABLE_RESULT_KEYS}.
The full bit depth of the sensor must be available in the raw data,
so the value for linear sensors should not be significantly lower
than maximum raw value supported, i.e. 2^(sensor bits per pixel).
n
3
Size in bytes for all the listed opaque RAW buffer sizes
Must be large enough to fit the opaque RAW of corresponding size produced by
the camera
This configurations are listed as `(width, height, size_in_bytes)` tuples.
This is used for sizing the gralloc buffers for opaque RAW buffers.
All RAW_OPAQUE output stream configuration listed in
android.scaler.availableStreamConfigurations will have a corresponding tuple in
this key.
This key is added in legacy HAL3.4.
For legacy HAL3.4 or above: devices advertising RAW_OPAQUE format output must list this
key. For legacy HAL3.3 or earlier devices: if RAW_OPAQUE ouput is advertised, camera
framework will derive this key by assuming each pixel takes two bytes and no padding bytes
between rows.
OFF
No lens shading correction is applied.
FAST
Apply lens shading corrections, without slowing
frame rate relative to sensor raw output
HIGH_QUALITY
Apply high-quality lens shading correction, at the
cost of possibly reduced frame rate.
Quality of lens shading correction applied
to the image data.
android.shading.availableModes
When set to OFF mode, no lens shading correction will be applied by the
camera device, and an identity lens shading map data will be provided
if `android.statistics.lensShadingMapMode == ON`. For example, for lens
shading map with size of `[ 4, 3 ]`,
the output android.statistics.lensShadingCorrectionMap for this case will be an identity
map shown below:
[ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ]
When set to other modes, lens shading correction will be applied by the camera
device. Applications can request lens shading map data by setting
android.statistics.lensShadingMapMode to ON, and then the camera device will provide lens
shading map data in android.statistics.lensShadingCorrectionMap; the returned shading map
data will be the one applied by the camera device for this capture request.
The shading map data may depend on the auto-exposure (AE) and AWB statistics, therefore
the reliability of the map data may be affected by the AE and AWB algorithms. When AE and
AWB are in AUTO modes(android.control.aeMode `!=` OFF and android.control.awbMode `!=`
OFF), to get best results, it is recommended that the applications wait for the AE and AWB
to be converged before using the returned shading map data.
Control the amount of shading correction
applied to the images
unitless: 1-10; 10 is full shading
compensation
n
List of lens shading modes for android.shading.mode that are supported by this camera device.
Any value listed in android.shading.mode
This list contains lens shading modes that can be set for the camera device.
Camera devices that support the MANUAL_POST_PROCESSING capability will always
list OFF and FAST mode. This includes all FULL level devices.
LEGACY devices will always only support FAST mode.
HAL must support both FAST and HIGH_QUALITY if lens shading correction control is
available on the camera device, but the underlying implementation can be the same for
both modes. That is, if the highest quality implementation on the camera device does not
slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.
OFF
Do not include face detection statistics in capture
results.
SIMPLE
Return face rectangle and confidence values only.
FULL
Return all face
metadata.
In this mode, face rectangles, scores, landmarks, and face IDs are all valid.
Operating mode for the face detector
unit.
android.statistics.info.availableFaceDetectModes
Whether face detection is enabled, and whether it
should output just the basic fields or the full set of
fields.
SIMPLE mode must fill in android.statistics.faceRectangles and
android.statistics.faceScores.
FULL mode must also fill in android.statistics.faceIds, and
android.statistics.faceLandmarks.
OFF
ON
Operating mode for histogram
generation
OFF
ON
Operating mode for sharpness map
generation
OFF
Hot pixel map production is disabled.
ON
Hot pixel map production is enabled.
Operating mode for hot pixel map generation.
android.statistics.info.availableHotPixelMapModes
If set to `true`, a hot pixel map is returned in android.statistics.hotPixelMap.
If set to `false`, no hot pixel map will be returned.
n
List of face detection modes for android.statistics.faceDetectMode that are
supported by this camera device.
Any value listed in android.statistics.faceDetectMode
OFF is always supported.
Number of histogram buckets
supported
>= 64
The maximum number of simultaneously detectable
faces.
0 for cameras without available face detection; otherwise:
`>=4` for LIMITED or FULL hwlevel devices or
`>0` for LEGACY devices.
Maximum value possible for a histogram
bucket
Maximum value possible for a sharpness map
region.
2
Dimensions of the sharpness
map
Must be at least 32 x 32
n
List of hot pixel map output modes for android.statistics.hotPixelMapMode that are
supported by this camera device.
Any value listed in android.statistics.hotPixelMapMode
If no hotpixel map output is available for this camera device, this will contain only
`false`.
ON is always supported on devices with the RAW capability.
n
List of lens shading map output modes for android.statistics.lensShadingMapMode that
are supported by this camera device.
Any value listed in android.statistics.lensShadingMapMode
If no lens shading map output is available for this camera device, this key will
contain only OFF.
ON is always supported on devices with the RAW capability.
LEGACY mode devices will always only support OFF.
n
List of OIS data output modes for android.statistics.oisDataMode that
are supported by this camera device.
Any value listed in android.statistics.oisDataMode
If no OIS data output is available for this camera device, this key will
contain only OFF.
n
List of unique IDs for detected faces.
Each detected face is given a unique ID that is valid for as long as the face is visible
to the camera device. A face that leaves the field of view and later returns may be
assigned a new ID.
Only available if android.statistics.faceDetectMode == FULL
n
6
List of landmarks for detected
faces.
For devices not supporting android.distortionCorrection.mode control, the coordinate
system always follows that of android.sensor.info.activeArraySize, with `(0, 0)` being
the top-left pixel of the active array.
For devices supporting android.distortionCorrection.mode control, the coordinate
system depends on the mode being set.
When the distortion correction mode is OFF, the coordinate system follows
android.sensor.info.preCorrectionActiveArraySize, with
`(0, 0)` being the top-left pixel of the pre-correction active array.
When the distortion correction mode is not OFF, the coordinate system follows
android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array.
Only available if android.statistics.faceDetectMode == FULL
HAL must always report face landmarks in the coordinate system of pre-correction
active array.
n
4
List of the bounding rectangles for detected
faces.
For devices not supporting android.distortionCorrection.mode control, the coordinate
system always follows that of android.sensor.info.activeArraySize, with `(0, 0)` being
the top-left pixel of the active array.
For devices supporting android.distortionCorrection.mode control, the coordinate
system depends on the mode being set.
When the distortion correction mode is OFF, the coordinate system follows
android.sensor.info.preCorrectionActiveArraySize, with
`(0, 0)` being the top-left pixel of the pre-correction active array.
When the distortion correction mode is not OFF, the coordinate system follows
android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array.
Only available if android.statistics.faceDetectMode != OFF
The data representation is `int[4]`, which maps to `(left, top, right, bottom)`.
HAL must always report face rectangles in the coordinate system of pre-correction
active array.
n
List of the face confidence scores for
detected faces
1-100
Only available if android.statistics.faceDetectMode != OFF.
The value should be meaningful (for example, setting 100 at
all times is illegal).
n
List of the faces detected through camera face detection
in this capture.
Only available if android.statistics.faceDetectMode `!=` OFF.
n
3
A 3-channel histogram based on the raw
sensor data
The k'th bucket (0-based) covers the input range
(with w = android.sensor.info.whiteLevel) of [ k * w/N,
(k + 1) * w / N ). If only a monochrome sharpness map is
supported, all channels should have the same data
n
m
3
A 3-channel sharpness map, based on the raw
sensor data
If only a monochrome sharpness map is supported,
all channels should have the same data
The shading map is a low-resolution floating-point map
that lists the coefficients used to correct for vignetting, for each
Bayer color channel.
Each gain factor is >= 1
The map provided here is the same map that is used by the camera device to
correct both color shading and vignetting for output non-RAW images.
When there is no lens shading correction applied to RAW
output images (android.sensor.info.lensShadingApplied `==`
false), this map is the complete lens shading correction
map; when there is some lens shading correction applied to
the RAW output image (android.sensor.info.lensShadingApplied
`==` true), this map reports the remaining lens shading
correction map that needs to be applied to get shading
corrected images that match the camera device's output for
non-RAW formats.
For a complete shading correction map, the least shaded
section of the image will have a gain factor of 1; all
other sections will have gains above 1.
When android.colorCorrection.mode = TRANSFORM_MATRIX, the map
will take into account the colorCorrection settings.
The shading map is for the entire active pixel array, and is not
affected by the crop region specified in the request. Each shading map
entry is the value of the shading compensation map over a specific
pixel on the sensor. Specifically, with a (N x M) resolution shading
map, and an active pixel array size (W x H), shading map entry
(x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at
pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels.
The map is assumed to be bilinearly interpolated between the sample points.
The channel order is [R, Geven, Godd, B], where Geven is the green
channel for the even rows of a Bayer pattern, and Godd is the odd rows.
The shading map is stored in a fully interleaved format.
The shading map will generally have on the order of 30-40 rows and columns,
and will be smaller than 64x64.
As an example, given a very small map defined as:
width,height = [ 4, 3 ]
values =
[ 1.3, 1.2, 1.15, 1.2, 1.2, 1.2, 1.15, 1.2,
1.1, 1.2, 1.2, 1.2, 1.3, 1.2, 1.3, 1.3,
1.2, 1.2, 1.25, 1.1, 1.1, 1.1, 1.1, 1.0,
1.0, 1.0, 1.0, 1.0, 1.2, 1.3, 1.25, 1.2,
1.3, 1.2, 1.2, 1.3, 1.2, 1.15, 1.1, 1.2,
1.2, 1.1, 1.0, 1.2, 1.3, 1.15, 1.2, 1.3 ]
The low-resolution scaling map images for each channel are
(displayed using nearest-neighbor interpolation):
As a visualization only, inverting the full-color map to recover an
image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives:
For a MONOCHROME camera, all of the 2x2 channels must have the same values. An example
shading map for such a camera is defined as:
android.lens.info.shadingMapSize = [ 4, 3 ]
android.statistics.lensShadingMap =
[ 1.3, 1.3, 1.3, 1.3, 1.2, 1.2, 1.2, 1.2,
1.1, 1.1, 1.1, 1.1, 1.3, 1.3, 1.3, 1.3,
1.2, 1.2, 1.2, 1.2, 1.1, 1.1, 1.1, 1.1,
1.0, 1.0, 1.0, 1.0, 1.2, 1.2, 1.2, 1.2,
1.3, 1.3, 1.3, 1.3, 1.2, 1.2, 1.2, 1.2,
1.2, 1.2, 1.2, 1.2, 1.3, 1.3, 1.3, 1.3 ]
4
n
m
The shading map is a low-resolution floating-point map
that lists the coefficients used to correct for vignetting and color shading,
for each Bayer color channel of RAW image data.
Each gain factor is >= 1
The map provided here is the same map that is used by the camera device to
correct both color shading and vignetting for output non-RAW images.
When there is no lens shading correction applied to RAW
output images (android.sensor.info.lensShadingApplied `==`
false), this map is the complete lens shading correction
map; when there is some lens shading correction applied to
the RAW output image (android.sensor.info.lensShadingApplied
`==` true), this map reports the remaining lens shading
correction map that needs to be applied to get shading
corrected images that match the camera device's output for
non-RAW formats.
For a complete shading correction map, the least shaded
section of the image will have a gain factor of 1; all
other sections will have gains above 1.
When android.colorCorrection.mode = TRANSFORM_MATRIX, the map
will take into account the colorCorrection settings.
The shading map is for the entire active pixel array, and is not
affected by the crop region specified in the request. Each shading map
entry is the value of the shading compensation map over a specific
pixel on the sensor. Specifically, with a (N x M) resolution shading
map, and an active pixel array size (W x H), shading map entry
(x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at
pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels.
The map is assumed to be bilinearly interpolated between the sample points.
For a Bayer camera, the channel order is [R, Geven, Godd, B], where Geven is
the green channel for the even rows of a Bayer pattern, and Godd is the odd rows.
The shading map is stored in a fully interleaved format, and its size
is provided in the camera static metadata by android.lens.info.shadingMapSize.
The shading map will generally have on the order of 30-40 rows and columns,
and will be smaller than 64x64.
As an example, given a very small map for a Bayer camera defined as:
android.lens.info.shadingMapSize = [ 4, 3 ]
android.statistics.lensShadingMap =
[ 1.3, 1.2, 1.15, 1.2, 1.2, 1.2, 1.15, 1.2,
1.1, 1.2, 1.2, 1.2, 1.3, 1.2, 1.3, 1.3,
1.2, 1.2, 1.25, 1.1, 1.1, 1.1, 1.1, 1.0,
1.0, 1.0, 1.0, 1.0, 1.2, 1.3, 1.25, 1.2,
1.3, 1.2, 1.2, 1.3, 1.2, 1.15, 1.1, 1.2,
1.2, 1.1, 1.0, 1.2, 1.3, 1.15, 1.2, 1.3 ]
The low-resolution scaling map images for each channel are
(displayed using nearest-neighbor interpolation):
As a visualization only, inverting the full-color map to recover an
image of a gray wall (using bicubic interpolation for visual quality)
as captured by the sensor gives:
For a MONOCHROME camera, all of the 2x2 channels must have the same values. An example
shading map for such a camera is defined as:
android.lens.info.shadingMapSize = [ 4, 3 ]
android.statistics.lensShadingMap =
[ 1.3, 1.3, 1.3, 1.3, 1.2, 1.2, 1.2, 1.2,
1.1, 1.1, 1.1, 1.1, 1.3, 1.3, 1.3, 1.3,
1.2, 1.2, 1.2, 1.2, 1.1, 1.1, 1.1, 1.1,
1.0, 1.0, 1.0, 1.0, 1.2, 1.2, 1.2, 1.2,
1.3, 1.3, 1.3, 1.3, 1.2, 1.2, 1.2, 1.2,
1.2, 1.2, 1.2, 1.2, 1.3, 1.3, 1.3, 1.3 ]
Note that the RAW image data might be subject to lens shading
correction not reported on this map. Query
android.sensor.info.lensShadingApplied to see if RAW image data has subject
to lens shading correction. If android.sensor.info.lensShadingApplied
is TRUE, the RAW image data is subject to partial or full lens shading
correction. In the case full lens shading correction is applied to RAW
images, the gain factor map reported in this key will contain all 1.0 gains.
In other words, the map reported in this key is the remaining lens shading
that needs to be applied on the RAW image to get images without lens shading
artifacts. See android.request.maxNumOutputRaw for a list of RAW image
formats.
The lens shading map calculation may depend on exposure and white balance statistics.
When AE and AWB are in AUTO modes
(android.control.aeMode `!=` OFF and android.control.awbMode `!=` OFF), the HAL
may have all the information it need to generate most accurate lens shading map. When
AE or AWB are in manual mode
(android.control.aeMode `==` OFF or android.control.awbMode `==` OFF), the shading map
may be adversely impacted by manual exposure or white balance parameters. To avoid
generating unreliable shading map data, the HAL may choose to lock the shading map with
the latest known good map generated when the AE and AWB are in AUTO modes.
4
The best-fit color channel gains calculated
by the camera device's statistics units for the current output frame.
Never fully implemented or specified; do not use
This may be different than the gains used for this frame,
since statistics processing on data from a new frame
typically completes after the transform has already been
applied to that frame.
The 4 channel gains are defined in Bayer domain,
see android.colorCorrection.gains for details.
This value should always be calculated by the auto-white balance (AWB) block,
regardless of the android.control.* current values.
3
3
The best-fit color transform matrix estimate
calculated by the camera device's statistics units for the current
output frame.
Never fully implemented or specified; do not use
The camera device will provide the estimate from its
statistics unit on the white balance transforms to use
for the next frame. These are the values the camera device believes
are the best fit for the current output frame. This may
be different than the transform used for this frame, since
statistics processing on data from a new frame typically
completes after the transform has already been applied to
that frame.
These estimates must be provided for all frames, even if
capture settings and color transforms are set by the application.
This value should always be calculated by the auto-white balance (AWB) block,
regardless of the android.control.* current values.
NONE
The camera device does not detect any flickering illumination
in the current scene.
50HZ
The camera device detects illumination flickering at 50Hz
in the current scene.
60HZ
The camera device detects illumination flickering at 60Hz
in the current scene.
The camera device estimated scene illumination lighting
frequency.
Many light sources, such as most fluorescent lights, flicker at a rate
that depends on the local utility power standards. This flicker must be
accounted for by auto-exposure routines to avoid artifacts in captured images.
The camera device uses this entry to tell the application what the scene
illuminant frequency is.
When manual exposure control is enabled
(`android.control.aeMode == OFF` or `android.control.mode ==
OFF`), the android.control.aeAntibandingMode doesn't perform
antibanding, and the application can ensure it selects
exposure times that do not cause banding issues by looking
into this metadata field. See
android.control.aeAntibandingMode for more details.
Reports NONE if there doesn't appear to be flickering illumination.
2
n
List of `(x, y)` coordinates of hot/defective pixels on the sensor.
n <= number of pixels on the sensor.
The `(x, y)` coordinates must be bounded by
android.sensor.info.pixelArraySize.
A coordinate `(x, y)` must lie between `(0, 0)`, and
`(width - 1, height - 1)` (inclusive), which are the top-left and
bottom-right of the pixel array, respectively. The width and
height dimensions are given in android.sensor.info.pixelArraySize.
This may include hot pixels that lie outside of the active array
bounds given by android.sensor.info.activeArraySize.
A hotpixel map contains the coordinates of pixels on the camera
sensor that do report valid values (usually due to defects in
the camera sensor). This includes pixels that are stuck at certain
values, or have a response that does not accuractly encode the
incoming light from the scene.
To avoid performance issues, there should be significantly fewer hot
pixels than actual pixels on the camera sensor.
OFF
Do not include a lens shading map in the capture result.
ON
Include a lens shading map in the capture result.
Whether the camera device will output the lens
shading map in output result metadata.
android.statistics.info.availableLensShadingMapModes
When set to ON,
android.statistics.lensShadingMap will be provided in
the output result metadata.
ON is always supported on devices with the RAW capability.
OFF
Do not include OIS data in the capture result.
ON
Include OIS data in the capture result.
android.statistics.oisSamples provides OIS sample data in the
output result metadata.
android.statistics.oisTimestamps, android.statistics.oisXShifts,
and android.statistics.oisYShifts provide OIS data in the output result metadata.
A control for selecting whether optical stabilization (OIS) position
information is included in output result metadata.
android.statistics.info.availableOisDataModes
Since optical image stabilization generally involves motion much faster than the duration
of individualq image exposure, multiple OIS samples can be included for a single capture
result. For example, if the OIS reporting operates at 200 Hz, a typical camera operating
at 30fps may have 6-7 OIS samples per capture result. This information can be combined
with the rolling shutter skew to account for lens motion during image exposure in
post-processing algorithms.
n
An array of timestamps of OIS samples, in nanoseconds.
nanoseconds
The array contains the timestamps of OIS samples. The timestamps are in the same
timebase as and comparable to android.sensor.timestamp.
n
An array of shifts of OIS samples, in x direction.
Pixels in active array.
The array contains the amount of shifts in x direction, in pixels, based on OIS samples.
A positive value is a shift from left to right in the pre-correction active array
coordinate system. For example, if the optical center is (1000, 500) in pre-correction
active array coordinates, a shift of (3, 0) puts the new optical center at (1003, 500).
The number of shifts must match the number of timestamps in
android.statistics.oisTimestamps.
The OIS samples are not affected by whether lens distortion correction is enabled (on
supporting devices). They are always reported in pre-correction active array coordinates,
since the scaling of OIS shifts would depend on the specific spot on the sensor the shift
is needed.
n
An array of shifts of OIS samples, in y direction.
Pixels in active array.
The array contains the amount of shifts in y direction, in pixels, based on OIS samples.
A positive value is a shift from top to bottom in pre-correction active array coordinate
system. For example, if the optical center is (1000, 500) in active array coordinates, a
shift of (0, 5) puts the new optical center at (1000, 505).
The number of shifts must match the number of timestamps in
android.statistics.oisTimestamps.
The OIS samples are not affected by whether lens distortion correction is enabled (on
supporting devices). They are always reported in pre-correction active array coordinates,
since the scaling of OIS shifts would depend on the specific spot on the sensor the shift
is needed.
n
An array of optical stabilization (OIS) position samples.
Each OIS sample contains the timestamp and the amount of shifts in x and y direction,
in pixels, of the OIS sample.
A positive value for a shift in x direction is a shift from left to right in the
pre-correction active array coordinate system. For example, if the optical center is
(1000, 500) in pre-correction active array coordinates, a shift of (3, 0) puts the new
optical center at (1003, 500).
A positive value for a shift in y direction is a shift from top to bottom in
pre-correction active array coordinate system. For example, if the optical center is
(1000, 500) in active array coordinates, a shift of (0, 5) puts the new optical center at
(1000, 505).
The OIS samples are not affected by whether lens distortion correction is enabled (on
supporting devices). They are always reported in pre-correction active array coordinates,
since the scaling of OIS shifts would depend on the specific spot on the sensor the shift
is needed.
n
2
Tonemapping / contrast / gamma curve for the blue
channel, to use when android.tonemap.mode is
CONTRAST_CURVE.
See android.tonemap.curveRed for more details.
n
2
Tonemapping / contrast / gamma curve for the green
channel, to use when android.tonemap.mode is
CONTRAST_CURVE.
See android.tonemap.curveRed for more details.
n
2
Tonemapping / contrast / gamma curve for the red
channel, to use when android.tonemap.mode is
CONTRAST_CURVE.
0-1 on both input and output coordinates, normalized
as a floating-point value such that 0 == black and 1 == white.
Each channel's curve is defined by an array of control points:
android.tonemap.curveRed =
[ P0in, P0out, P1in, P1out, P2in, P2out, P3in, P3out, ..., PNin, PNout ]
2 <= N <= android.tonemap.maxCurvePoints
These are sorted in order of increasing `Pin`; it is
required that input values 0.0 and 1.0 are included in the list to
define a complete mapping. For input values between control points,
the camera device must linearly interpolate between the control
points.
Each curve can have an independent number of points, and the number
of points can be less than max (that is, the request doesn't have to
always provide a curve with number of points equivalent to
android.tonemap.maxCurvePoints).
For devices with MONOCHROME capability, all three channels must have the same set of
control points.
A few examples, and their corresponding graphical mappings; these
only specify the red channel and the precision is limited to 4
digits, for conciseness.
Linear mapping:
android.tonemap.curveRed = [ 0, 0, 1.0, 1.0 ]
Invert mapping:
android.tonemap.curveRed = [ 0, 1.0, 1.0, 0 ]
Gamma 1/2.2 mapping, with 16 control points:
android.tonemap.curveRed = [
0.0000, 0.0000, 0.0667, 0.2920, 0.1333, 0.4002, 0.2000, 0.4812,
0.2667, 0.5484, 0.3333, 0.6069, 0.4000, 0.6594, 0.4667, 0.7072,
0.5333, 0.7515, 0.6000, 0.7928, 0.6667, 0.8317, 0.7333, 0.8685,
0.8000, 0.9035, 0.8667, 0.9370, 0.9333, 0.9691, 1.0000, 1.0000 ]
Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:
android.tonemap.curveRed = [
0.0000, 0.0000, 0.0667, 0.2864, 0.1333, 0.4007, 0.2000, 0.4845,
0.2667, 0.5532, 0.3333, 0.6125, 0.4000, 0.6652, 0.4667, 0.7130,
0.5333, 0.7569, 0.6000, 0.7977, 0.6667, 0.8360, 0.7333, 0.8721,
0.8000, 0.9063, 0.8667, 0.9389, 0.9333, 0.9701, 1.0000, 1.0000 ]
For good quality of mapping, at least 128 control points are
preferred.
A typical use case of this would be a gamma-1/2.2 curve, with as many
control points used as are available.
Tonemapping / contrast / gamma curve to use when android.tonemap.mode
is CONTRAST_CURVE.
The tonemapCurve consist of three curves for each of red, green, and blue
channels respectively. The following example uses the red channel as an
example. The same logic applies to green and blue channel.
Each channel's curve is defined by an array of control points:
curveRed =
[ P0(in, out), P1(in, out), P2(in, out), P3(in, out), ..., PN(in, out) ]
2 <= N <= android.tonemap.maxCurvePoints
These are sorted in order of increasing `Pin`; it is always
guaranteed that input values 0.0 and 1.0 are included in the list to
define a complete mapping. For input values between control points,
the camera device must linearly interpolate between the control
points.
Each curve can have an independent number of points, and the number
of points can be less than max (that is, the request doesn't have to
always provide a curve with number of points equivalent to
android.tonemap.maxCurvePoints).
For devices with MONOCHROME capability, all three channels must have the same set of
control points.
A few examples, and their corresponding graphical mappings; these
only specify the red channel and the precision is limited to 4
digits, for conciseness.
Linear mapping:
curveRed = [ (0, 0), (1.0, 1.0) ]
Invert mapping:
curveRed = [ (0, 1.0), (1.0, 0) ]
Gamma 1/2.2 mapping, with 16 control points:
curveRed = [
(0.0000, 0.0000), (0.0667, 0.2920), (0.1333, 0.4002), (0.2000, 0.4812),
(0.2667, 0.5484), (0.3333, 0.6069), (0.4000, 0.6594), (0.4667, 0.7072),
(0.5333, 0.7515), (0.6000, 0.7928), (0.6667, 0.8317), (0.7333, 0.8685),
(0.8000, 0.9035), (0.8667, 0.9370), (0.9333, 0.9691), (1.0000, 1.0000) ]
Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:
curveRed = [
(0.0000, 0.0000), (0.0667, 0.2864), (0.1333, 0.4007), (0.2000, 0.4845),
(0.2667, 0.5532), (0.3333, 0.6125), (0.4000, 0.6652), (0.4667, 0.7130),
(0.5333, 0.7569), (0.6000, 0.7977), (0.6667, 0.8360), (0.7333, 0.8721),
(0.8000, 0.9063), (0.8667, 0.9389), (0.9333, 0.9701), (1.0000, 1.0000) ]
This entry is created by the framework from the curveRed, curveGreen and
curveBlue entries.
CONTRAST_CURVE
Use the tone mapping curve specified in
the android.tonemap.curve* entries.
All color enhancement and tonemapping must be disabled, except
for applying the tonemapping curve specified by
android.tonemap.curve.
Must not slow down frame rate relative to raw
sensor output.
FAST
Advanced gamma mapping and color enhancement may be applied, without
reducing frame rate compared to raw sensor output.
HIGH_QUALITY
High-quality gamma mapping and color enhancement will be applied, at
the cost of possibly reduced frame rate compared to raw sensor output.
GAMMA_VALUE
Use the gamma value specified in android.tonemap.gamma to peform
tonemapping.
All color enhancement and tonemapping must be disabled, except
for applying the tonemapping curve specified by android.tonemap.gamma.
Must not slow down frame rate relative to raw sensor output.
PRESET_CURVE
Use the preset tonemapping curve specified in
android.tonemap.presetCurve to peform tonemapping.
All color enhancement and tonemapping must be disabled, except
for applying the tonemapping curve specified by
android.tonemap.presetCurve.
Must not slow down frame rate relative to raw sensor output.
High-level global contrast/gamma/tonemapping control.
android.tonemap.availableToneMapModes
When switching to an application-defined contrast curve by setting
android.tonemap.mode to CONTRAST_CURVE, the curve is defined
per-channel with a set of `(in, out)` points that specify the
mapping from input high-bit-depth pixel value to the output
low-bit-depth value. Since the actual pixel ranges of both input
and output may change depending on the camera pipeline, the values
are specified by normalized floating-point numbers.
More-complex color mapping operations such as 3D color look-up
tables, selective chroma enhancement, or other non-linear color
transforms will be disabled when android.tonemap.mode is
CONTRAST_CURVE.
When using either FAST or HIGH_QUALITY, the camera device will
emit its own tonemap curve in android.tonemap.curve.
These values are always available, and as close as possible to the
actually used nonlinear/nonglobal transforms.
If a request is sent with CONTRAST_CURVE with the camera device's
provided curve in FAST or HIGH_QUALITY, the image's tonemap will be
roughly the same.
Maximum number of supported points in the
tonemap curve that can be used for android.tonemap.curve.
If the actual number of points provided by the application (in android.tonemap.curve*) is
less than this maximum, the camera device will resample the curve to its internal
representation, using linear interpolation.
The output curves in the result metadata may have a different number
of points than the input curves, and will represent the actual
hardware curves used as closely as possible when linearly interpolated.
This value must be at least 64. This should be at least 128.
n
List of tonemapping modes for android.tonemap.mode that are supported by this camera
device.
Any value listed in android.tonemap.mode
Camera devices that support the MANUAL_POST_PROCESSING capability will always contain
at least one of below mode combinations:
* CONTRAST_CURVE, FAST and HIGH_QUALITY
* GAMMA_VALUE, PRESET_CURVE, FAST and HIGH_QUALITY
This includes all FULL level devices.
HAL must support both FAST and HIGH_QUALITY if automatic tonemap control is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY will generate the same output.
Tonemapping curve to use when android.tonemap.mode is
GAMMA_VALUE
The tonemap curve will be defined the following formula:
* OUT = pow(IN, 1.0 / gamma)
where IN and OUT is the input pixel value scaled to range [0.0, 1.0],
pow is the power function and gamma is the gamma value specified by this
key.
The same curve will be applied to all color channels. The camera device
may clip the input gamma value to its supported range. The actual applied
value will be returned in capture result.
The valid range of gamma value varies on different devices, but values
within [1.0, 5.0] are guaranteed not to be clipped.
SRGB
Tonemapping curve is defined by sRGB
REC709
Tonemapping curve is defined by ITU-R BT.709
Tonemapping curve to use when android.tonemap.mode is
PRESET_CURVE
The tonemap curve will be defined by specified standard.
sRGB (approximated by 16 control points):
Rec. 709 (approximated by 16 control points):
Note that above figures show a 16 control points approximation of preset
curves. Camera devices may apply a different approximation to the curve.
OFF
ON
This LED is nominally used to indicate to the user
that the camera is powered on and may be streaming images back to the
Application Processor. In certain rare circumstances, the OS may
disable this when video is processed locally and not transmitted to
any untrusted applications.
In particular, the LED *must* always be on when the data could be
transmitted off the device. The LED *should* always be on whenever
data is stored locally on the device.
The LED *may* be off if a trusted application is using the data that
doesn't violate the above rules.
n
TRANSMIT
android.led.transmit control is used.
A list of camera LEDs that are available on this system.
LIMITED
This camera device does not have enough capabilities to qualify as a `FULL` device or
better.
Only the stream configurations listed in the `LEGACY` and `LIMITED` tables in the
{@link android.hardware.camera2.CameraDevice#createCaptureSession|ACameraDevice_createCaptureSession
createCaptureSession} documentation are guaranteed to be supported.
All `LIMITED` devices support the `BACKWARDS_COMPATIBLE` capability, indicating basic
support for color image capture. The only exception is that the device may
alternatively support only the `DEPTH_OUTPUT` capability, if it can only output depth
measurements and not color images.
`LIMITED` devices and above require the use of android.control.aePrecaptureTrigger
to lock exposure metering (and calculate flash power, for cameras with flash) before
capturing a high-quality still image.
A `LIMITED` device that only lists the `BACKWARDS_COMPATIBLE` capability is only
required to support full-automatic operation and post-processing (`OFF` is not
supported for android.control.aeMode, android.control.afMode, or
android.control.awbMode)
Additional capabilities may optionally be supported by a `LIMITED`-level device, and
can be checked for in android.request.availableCapabilities.
FULL
This camera device is capable of supporting advanced imaging applications.
The stream configurations listed in the `FULL`, `LEGACY` and `LIMITED` tables in the
{@link android.hardware.camera2.CameraDevice#createCaptureSession|ACameraDevice_createCaptureSession
createCaptureSession} documentation are guaranteed to be supported.
A `FULL` device will support below capabilities:
* `BURST_CAPTURE` capability (android.request.availableCapabilities contains
`BURST_CAPTURE`)
* Per frame control (android.sync.maxLatency `==` PER_FRAME_CONTROL)
* Manual sensor control (android.request.availableCapabilities contains `MANUAL_SENSOR`)
* Manual post-processing control (android.request.availableCapabilities contains
`MANUAL_POST_PROCESSING`)
* The required exposure time range defined in android.sensor.info.exposureTimeRange
* The required maxFrameDuration defined in android.sensor.info.maxFrameDuration
Note:
Pre-API level 23, FULL devices also supported arbitrary cropping region
(android.scaler.croppingType `== FREEFORM`); this requirement was relaxed in API level
23, and `FULL` devices may only support `CENTERED` cropping.
LEGACY
This camera device is running in backward compatibility mode.
Only the stream configurations listed in the `LEGACY` table in the {@link
android.hardware.camera2.CameraDevice#createCaptureSession|ACameraDevice_createCaptureSession
createCaptureSession} documentation are supported.
A `LEGACY` device does not support per-frame control, manual sensor control, manual
post-processing, arbitrary cropping regions, and has relaxed performance constraints.
No additional capabilities beyond `BACKWARD_COMPATIBLE` will ever be listed by a
`LEGACY` device in android.request.availableCapabilities.
In addition, the android.control.aePrecaptureTrigger is not functional on `LEGACY`
devices. Instead, every request that includes a JPEG-format output target is treated
as triggering a still capture, internally executing a precapture trigger. This may
fire the flash for flash power metering during precapture, and then fire the flash
for the final capture, if a flash is available on the device and the AE mode is set to
enable the flash.
Devices that initially shipped with Android version {@link
android.os.Build.VERSION_CODES#Q Q} or newer will not include any LEGACY-level devices.
3
This camera device is capable of YUV reprocessing and RAW data capture, in addition to
FULL-level capabilities.
The stream configurations listed in the `LEVEL_3`, `RAW`, `FULL`, `LEGACY` and
`LIMITED` tables in the {@link
android.hardware.camera2.CameraDevice#createCaptureSession|ACameraDevice_createCaptureSession
createCaptureSession} documentation are guaranteed to be supported.
The following additional capabilities are guaranteed to be supported:
* `YUV_REPROCESSING` capability (android.request.availableCapabilities contains
`YUV_REPROCESSING`)
* `RAW` capability (android.request.availableCapabilities contains
`RAW`)
EXTERNAL
This camera device is backed by an external camera connected to this Android device.
The device has capability identical to a LIMITED level device, with the following
exceptions:
* The device may not report lens/sensor related information such as
- android.lens.focalLength
- android.lens.info.hyperfocalDistance
- android.sensor.info.physicalSize
- android.sensor.info.whiteLevel
- android.sensor.blackLevelPattern
- android.sensor.info.colorFilterArrangement
- android.sensor.rollingShutterSkew
* The device will report 0 for android.sensor.orientation
* The device has less guarantee on stable framerate, as the framerate partly depends
on the external camera being used.
Generally classifies the overall set of the camera device functionality.
The supported hardware level is a high-level description of the camera device's
capabilities, summarizing several capabilities into one field. Each level adds additional
features to the previous one, and is always a strict superset of the previous level.
The ordering is `LEGACY < LIMITED < FULL < LEVEL_3`.
Starting from `LEVEL_3`, the level enumerations are guaranteed to be in increasing
numerical value as well. To check if a given device is at least at a given hardware level,
the following code snippet can be used:
// Returns true if the device supports the required hardware level, or better.
boolean isHardwareLevelSupported(CameraCharacteristics c, int requiredLevel) {
final int[] sortedHwLevels = {
CameraCharacteristics.INFO_SUPPORTED_HARDWARE_LEVEL_LEGACY,
CameraCharacteristics.INFO_SUPPORTED_HARDWARE_LEVEL_EXTERNAL,
CameraCharacteristics.INFO_SUPPORTED_HARDWARE_LEVEL_LIMITED,
CameraCharacteristics.INFO_SUPPORTED_HARDWARE_LEVEL_FULL,
CameraCharacteristics.INFO_SUPPORTED_HARDWARE_LEVEL_3
};
int deviceLevel = c.get(CameraCharacteristics.INFO_SUPPORTED_HARDWARE_LEVEL);
if (requiredLevel == deviceLevel) {
return true;
}
for (int sortedlevel : sortedHwLevels) {
if (sortedlevel == requiredLevel) {
return true;
} else if (sortedlevel == deviceLevel) {
return false;
}
}
return false; // Should never reach here
}
At a high level, the levels are:
* `LEGACY` devices operate in a backwards-compatibility mode for older
Android devices, and have very limited capabilities.
* `LIMITED` devices represent the
baseline feature set, and may also include additional capabilities that are
subsets of `FULL`.
* `FULL` devices additionally support per-frame manual control of sensor, flash, lens and
post-processing settings, and image capture at a high rate.
* `LEVEL_3` devices additionally support YUV reprocessing and RAW image capture, along
with additional output stream configurations.
* `EXTERNAL` devices are similar to `LIMITED` devices with exceptions like some sensor or
lens information not reported or less stable framerates.
See the individual level enums for full descriptions of the supported capabilities. The
android.request.availableCapabilities entry describes the device's capabilities at a
finer-grain level, if needed. In addition, many controls have their available settings or
ranges defined in individual entries from {@link
android.hardware.camera2.CameraCharacteristics|ACameraManager_getCameraCharacteristics}.
Some features are not part of any particular hardware level or capability and must be
queried separately. These include:
* Calibrated timestamps (android.sensor.info.timestampSource `==` REALTIME)
* Precision lens control (android.lens.info.focusDistanceCalibration `==` CALIBRATED)
* Face detection (android.statistics.info.availableFaceDetectModes)
* Optical or electrical image stabilization
(android.lens.info.availableOpticalStabilization,
android.control.availableVideoStabilizationModes)
A camera HALv3 device can implement one of three possible operational modes; LIMITED,
FULL, and LEVEL_3.
FULL support or better is expected from new higher-end devices. Limited
mode has hardware requirements roughly in line with those for a camera HAL device v1
implementation, and is expected from older or inexpensive devices. Each level is a strict
superset of the previous level, and they share the same essential operational flow.
For full details refer to "S3. Operational Modes" in camera3.h
Camera HAL3+ must not implement LEGACY mode. It is there for backwards compatibility in
the `android.hardware.camera2` user-facing API only on legacy HALv1 devices, and is
implemented by the camera framework code.
EXTERNAL level devices have lower peformance bar in CTS since the peformance might depend
on the external camera being used and is not fully controlled by the device manufacturer.
The ITS test suite is exempted for the same reason.
A short string for manufacturer version information about the camera device, such as
ISP hardware, sensors, etc.
This can be used in {@link android.media.ExifInterface#TAG_IMAGE_DESCRIPTION TAG_IMAGE_DESCRIPTION}
in jpeg EXIF. This key may be absent if no version information is available on the
device.
The string must consist of only alphanumeric characters, punctuation, and
whitespace, i.e. it must match regular expression "[\p{Alnum}\p{Punct}\p{Space}]*".
It must not exceed 256 characters.
HIDL_DEVICE_3_5
This camera device supports and opts in to the buffer management APIs provided by
HIDL ICameraDevice version 3.5.
The version of buffer management API this camera device supports and opts into.
When this key is not present, camera framework will interact with this camera device
without any buffer management HAL API. When this key is present and camera framework
supports the buffer management API version, camera framework will interact with camera
HAL using such version of buffer management API.
OFF
ON
Whether black-level compensation is locked
to its current values, or is free to vary.
When set to `true` (ON), the values used for black-level
compensation will not change until the lock is set to
`false` (OFF).
Since changes to certain capture parameters (such as
exposure time) may require resetting of black level
compensation, the camera device must report whether setting
the black level lock was successful in the output result
metadata.
For example, if a sequence of requests is as follows:
* Request 1: Exposure = 10ms, Black level lock = OFF
* Request 2: Exposure = 10ms, Black level lock = ON
* Request 3: Exposure = 10ms, Black level lock = ON
* Request 4: Exposure = 20ms, Black level lock = ON
* Request 5: Exposure = 20ms, Black level lock = ON
* Request 6: Exposure = 20ms, Black level lock = ON
And the exposure change in Request 4 requires the camera
device to reset the black level offsets, then the output
result metadata is expected to be:
* Result 1: Exposure = 10ms, Black level lock = OFF
* Result 2: Exposure = 10ms, Black level lock = ON
* Result 3: Exposure = 10ms, Black level lock = ON
* Result 4: Exposure = 20ms, Black level lock = OFF
* Result 5: Exposure = 20ms, Black level lock = ON
* Result 6: Exposure = 20ms, Black level lock = ON
This indicates to the application that on frame 4, black
levels were reset due to exposure value changes, and pixel
values may not be consistent across captures.
The camera device will maintain the lock to the extent
possible, only overriding the lock to OFF when changes to
other request parameters require a black level recalculation
or reset.
If for some reason black level locking is no longer possible
(for example, the analog gain has changed, which forces
black level offsets to be recalculated), then the HAL must
override this request (and it must report 'OFF' when this
does happen) until the next capture for which locking is
possible again.
Whether the black level offset was locked for this frame. Should be
ON if android.blackLevel.lock was ON in the capture request, unless
a change in other capture settings forced the camera device to
perform a black level reset.
CONVERGING
The current result is not yet fully synchronized to any request.
Synchronization is in progress, and reading metadata from this
result may include a mix of data that have taken effect since the
last synchronization time.
In some future result, within android.sync.maxLatency frames,
this value will update to the actual frame number frame number
the result is guaranteed to be synchronized to (as long as the
request settings remain constant).
UNKNOWN
The current result's synchronization status is unknown.
The result may have already converged, or it may be in
progress. Reading from this result may include some mix
of settings from past requests.
After a settings change, the new settings will eventually all
take effect for the output buffers and results. However, this
value will not change when that happens. Altering settings
rapidly may provide outcomes using mixes of settings from recent
requests.
This value is intended primarily for backwards compatibility with
the older camera implementations (for android.hardware.Camera).
The frame number corresponding to the last request
with which the output result (metadata + buffers) has been fully
synchronized.
Either a non-negative value corresponding to a
`frame_number`, or one of the two enums (CONVERGING / UNKNOWN).
When a request is submitted to the camera device, there is usually a
delay of several frames before the controls get applied. A camera
device may either choose to account for this delay by implementing a
pipeline and carefully submit well-timed atomic control updates, or
it may start streaming control changes that span over several frame
boundaries.
In the latter case, whenever a request's settings change relative to
the previous submitted request, the full set of changes may take
multiple frame durations to fully take effect. Some settings may
take effect sooner (in less frame durations) than others.
While a set of control changes are being propagated, this value
will be CONVERGING.
Once it is fully known that a set of control changes have been
finished propagating, and the resulting updated control settings
have been read back by the camera device, this value will be set
to a non-negative frame number (corresponding to the request to
which the results have synchronized to).
Older camera device implementations may not have a way to detect
when all camera controls have been applied, and will always set this
value to UNKNOWN.
FULL capability devices will always have this value set to the
frame number of the request corresponding to this result.
_Further details_:
* Whenever a request differs from the last request, any future
results not yet returned may have this value set to CONVERGING (this
could include any in-progress captures not yet returned by the camera
device, for more details see pipeline considerations below).
* Submitting a series of multiple requests that differ from the
previous request (e.g. r1, r2, r3 s.t. r1 != r2 != r3)
moves the new synchronization frame to the last non-repeating
request (using the smallest frame number from the contiguous list of
repeating requests).
* Submitting the same request repeatedly will not change this value
to CONVERGING, if it was already a non-negative value.
* When this value changes to non-negative, that means that all of the
metadata controls from the request have been applied, all of the
metadata controls from the camera device have been read to the
updated values (into the result), and all of the graphics buffers
corresponding to this result are also synchronized to the request.
_Pipeline considerations_:
Submitting a request with updated controls relative to the previously
submitted requests may also invalidate the synchronization state
of all the results corresponding to currently in-flight requests.
In other words, results for this current request and up to
android.request.pipelineMaxDepth prior requests may have their
android.sync.frameNumber change to CONVERGING.
Using UNKNOWN here is illegal unless android.sync.maxLatency
is also UNKNOWN.
FULL capability devices should simply set this value to the
`frame_number` of the request this result corresponds to.
PER_FRAME_CONTROL
Every frame has the requests immediately applied.
Changing controls over multiple requests one after another will
produce results that have those controls applied atomically
each frame.
All FULL capability devices will have this as their maxLatency.
UNKNOWN
Each new frame has some subset (potentially the entire set)
of the past requests applied to the camera settings.
By submitting a series of identical requests, the camera device
will eventually have the camera settings applied, but it is
unknown when that exact point will be.
All LEGACY capability devices will have this as their maxLatency.
The maximum number of frames that can occur after a request
(different than the previous) has been submitted, and before the
result's state becomes synchronized.
Frame counts
A positive value, PER_FRAME_CONTROL, or UNKNOWN.
This defines the maximum distance (in number of metadata results),
between the frame number of the request that has new controls to apply
and the frame number of the result that has all the controls applied.
In other words this acts as an upper boundary for how many frames
must occur before the camera device knows for a fact that the new
submitted camera settings have been applied in outgoing frames.
For example if maxLatency was 2,
initial request = X (repeating)
request1 = X
request2 = Y
request3 = Y
request4 = Y
where requestN has frameNumber N, and the first of the repeating
initial request's has frameNumber F (and F < 1).
initial result = X' + { android.sync.frameNumber == F }
result1 = X' + { android.sync.frameNumber == F }
result2 = X' + { android.sync.frameNumber == CONVERGING }
result3 = X' + { android.sync.frameNumber == CONVERGING }
result4 = X' + { android.sync.frameNumber == 2 }
where resultN has frameNumber N.
Since `result4` has a `frameNumber == 4` and
`android.sync.frameNumber == 2`, the distance is clearly
`4 - 2 = 2`.
Use `frame_count` from camera3_request_t instead of
android.request.frameCount or
`{@link android.hardware.camera2.CaptureResult#getFrameNumber}`.
LIMITED devices are strongly encouraged to use a non-negative
value. If UNKNOWN is used here then app developers do not have a way
to know when sensor settings have been applied.
The amount of exposure time increase factor applied to the original output
frame by the application processing before sending for reprocessing.
Relative exposure time increase factor.
>= 1.0
This is optional, and will be supported if the camera device supports YUV_REPROCESSING
capability (android.request.availableCapabilities contains YUV_REPROCESSING).
For some YUV reprocessing use cases, the application may choose to filter the original
output frames to effectively reduce the noise to the same level as a frame that was
captured with longer exposure time. To be more specific, assuming the original captured
images were captured with a sensitivity of S and an exposure time of T, the model in
the camera device is that the amount of noise in the image would be approximately what
would be expected if the original capture parameters had been a sensitivity of
S/effectiveExposureFactor and an exposure time of T*effectiveExposureFactor, rather
than S and T respectively. If the captured images were processed by the application
before being sent for reprocessing, then the application may have used image processing
algorithms and/or multi-frame image fusion to reduce the noise in the
application-processed images (input images). By using the effectiveExposureFactor
control, the application can communicate to the camera device the actual noise level
improvement in the application-processed image. With this information, the camera
device can select appropriate noise reduction and edge enhancement parameters to avoid
excessive noise reduction (android.noiseReduction.mode) and insufficient edge
enhancement (android.edge.mode) being applied to the reprocessed frames.
For example, for multi-frame image fusion use case, the application may fuse
multiple output frames together to a final frame for reprocessing. When N image are
fused into 1 image for reprocessing, the exposure time increase factor could be up to
square root of N (based on a simple photon shot noise model). The camera device will
adjust the reprocessing noise reduction and edge enhancement parameters accordingly to
produce the best quality images.
This is relative factor, 1.0 indicates the application hasn't processed the input
buffer in a way that affects its effective exposure time.
This control is only effective for YUV reprocessing capture request. For noise
reduction reprocessing, it is only effective when `android.noiseReduction.mode != OFF`.
Similarly, for edge enhancement reprocessing, it is only effective when
`android.edge.mode != OFF`.
The maximal camera capture pipeline stall (in unit of frame count) introduced by a
reprocess capture request.
Number of frames.
<= 4
The key describes the maximal interference that one reprocess (input) request
can introduce to the camera simultaneous streaming of regular (output) capture
requests, including repeating requests.
When a reprocessing capture request is submitted while a camera output repeating request
(e.g. preview) is being served by the camera device, it may preempt the camera capture
pipeline for at least one frame duration so that the camera device is unable to process
the following capture request in time for the next sensor start of exposure boundary.
When this happens, the application may observe a capture time gap (longer than one frame
duration) between adjacent capture output frames, which usually exhibits as preview
glitch if the repeating request output targets include a preview surface. This key gives
the worst-case number of frame stall introduced by one reprocess request with any kind of
formats/sizes combination.
If this key reports 0, it means a reprocess request doesn't introduce any glitch to the
ongoing camera repeating request outputs, as if this reprocess request is never issued.
This key is supported if the camera device supports PRIVATE or YUV reprocessing (
i.e. android.request.availableCapabilities contains PRIVATE_REPROCESSING or
YUV_REPROCESSING).
Maximum number of points that a depth point cloud may contain.
If a camera device supports outputting depth range data in the form of a depth point
cloud ({@link android.graphics.ImageFormat#DEPTH_POINT_CLOUD}), this is the maximum
number of points an output buffer may contain.
Any given buffer may contain between 0 and maxDepthSamples points, inclusive.
If output in the depth point cloud format is not supported, this entry will
not be defined.
n
4
OUTPUT
INPUT
The available depth dataspace stream
configurations that this camera device supports
(i.e. format, width, height, output/input stream).
These are output stream configurations for use with
dataSpace HAL_DATASPACE_DEPTH. The configurations are
listed as `(format, width, height, input?)` tuples.
Only devices that support depth output for at least
the HAL_PIXEL_FORMAT_Y16 dense depth map may include
this entry.
A device that also supports the HAL_PIXEL_FORMAT_BLOB
sparse depth point cloud must report a single entry for
the format in this list as `(HAL_PIXEL_FORMAT_BLOB,
android.depth.maxDepthSamples, 1, OUTPUT)` in addition to
the entries for HAL_PIXEL_FORMAT_Y16.
4
n
This lists the minimum frame duration for each
format/size combination for depth output formats.
(format, width, height, ns) x n
This should correspond to the frame duration when only that
stream is active, with all processing (typically in android.*.mode)
set to either OFF or FAST.
When multiple streams are used in a request, the minimum frame
duration will be max(individual stream min durations).
The minimum frame duration of a stream (of a particular format, size)
is the same regardless of whether the stream is input or output.
See android.sensor.frameDuration and
android.scaler.availableStallDurations for more details about
calculating the max frame rate.
4
n
This lists the maximum stall duration for each
output format/size combination for depth streams.
(format, width, height, ns) x n
A stall duration is how much extra time would get added
to the normal minimum frame duration for a repeating request
that has streams with non-zero stall.
This functions similarly to
android.scaler.availableStallDurations for depth
streams.
All depth output stream formats may have a nonzero stall
duration.
FALSE
TRUE
Indicates whether a capture request may target both a
DEPTH16 / DEPTH_POINT_CLOUD output, and normal color outputs (such as
YUV_420_888, JPEG, or RAW) simultaneously.
If TRUE, including both depth and color outputs in a single
capture request is not supported. An application must interleave color
and depth requests. If FALSE, a single request can target both types
of output.
Typically, this restriction exists on camera devices that
need to emit a specific pattern or wavelength of light to
measure depth values, which causes the color image to be
corrupted during depth measurement.
n
5
Recommended depth stream configurations for common client use cases.
Optional subset of the android.depth.availableDepthStreamConfigurations that
contains similar tuples listed as
(i.e. width, height, format, output/input stream, usecase bit field).
Camera devices will be able to suggest particular depth stream configurations which are
power and performance efficient for specific use cases. For more information about
retrieving the suggestions see
{@link android.hardware.camera2.CameraCharacteristics#getRecommendedStreamConfigurationMap}.
For data representation please refer to
android.scaler.availableRecommendedStreamConfigurations
Recommended depth configurations are expected to be declared with SNAPSHOT and/or
ZSL if supported by the device.
For additional details on how to declare recommended stream configurations, check
android.scaler.availableRecommendedStreamConfigurations.
For additional requirements on depth streams please consider
android.depth.availableDepthStreamConfigurations.
n
4
OUTPUT
INPUT
The available dynamic depth dataspace stream
configurations that this camera device supports
(i.e. format, width, height, output/input stream).
These are output stream configurations for use with
dataSpace DYNAMIC_DEPTH. The configurations are
listed as `(format, width, height, input?)` tuples.
Only devices that support depth output for at least
the HAL_PIXEL_FORMAT_Y16 dense depth map along with
HAL_PIXEL_FORMAT_BLOB with the same size or size with
the same aspect ratio can have dynamic depth dataspace
stream configuration. android.depth.depthIsExclusive also
needs to be set to FALSE.
Do not set this property directly.
It is populated by camera framework and must not be set
at the HAL layer.
4
n
This lists the minimum frame duration for each
format/size combination for dynamic depth output streams.
(format, width, height, ns) x n
This should correspond to the frame duration when only that
stream is active, with all processing (typically in android.*.mode)
set to either OFF or FAST.
When multiple streams are used in a request, the minimum frame
duration will be max(individual stream min durations).
The minimum frame duration of a stream (of a particular format, size)
is the same regardless of whether the stream is input or output.
Do not set this property directly.
It is populated by camera framework and must not be set
at the HAL layer.
4
n
This lists the maximum stall duration for each
output format/size combination for dynamic depth streams.
(format, width, height, ns) x n
A stall duration is how much extra time would get added
to the normal minimum frame duration for a repeating request
that has streams with non-zero stall.
All dynamic depth output streams may have a nonzero stall
duration.
Do not set this property directly.
It is populated by camera framework and must not be set
at the HAL layer.
n
String containing the ids of the underlying physical cameras.
UTF-8 null-terminated string
For a logical camera, this is concatenation of all underlying physical camera IDs.
The null terminator for physical camera ID must be preserved so that the whole string
can be tokenized using '\0' to generate list of physical camera IDs.
For example, if the physical camera IDs of the logical camera are "2" and "3", the
value of this tag will be ['2', '\0', '3', '\0'].
The number of physical camera IDs must be no less than 2.
APPROXIMATE
A software mechanism is used to synchronize between the physical cameras. As a result,
the timestamp of an image from a physical stream is only an approximation of the
image sensor start-of-exposure time.
CALIBRATED
The camera device supports frame timestamp synchronization at the hardware level,
and the timestamp of a physical stream image accurately reflects its
start-of-exposure time.
The accuracy of frame timestamp synchronization between physical cameras
The accuracy of the frame timestamp synchronization determines the physical cameras'
ability to start exposure at the same time. If the sensorSyncType is CALIBRATED,
the physical camera sensors usually run in master-slave mode so that their shutter
time is synchronized. For APPROXIMATE sensorSyncType, the camera sensors usually run in
master-master mode, and there could be offset between their start of exposure.
In both cases, all images generated for a particular capture request still carry the same
timestamps, so that they can be used to look up the matching frame number and
onCaptureStarted callback.
This tag is only applicable if the logical camera device supports concurrent physical
streams from different physical cameras.
String containing the ID of the underlying active physical camera.
UTF-8 null-terminated string
The ID of the active physical camera that's backing the logical camera. All camera
streams and metadata that are not physical camera specific will be originating from this
physical camera.
For a logical camera made up of physical cameras where each camera's lenses have
different characteristics, the camera device may choose to switch between the physical
cameras when application changes FOCAL_LENGTH or SCALER_CROP_REGION.
At the time of lens switch, this result metadata reflects the new active physical camera
ID.
This key will be available if the camera device advertises this key via {@link
android.hardware.camera2.CameraCharacteristics#getAvailableCaptureResultKeys|ACAMERA_REQUEST_AVAILABLE_RESULT_KEYS}.
When available, this must be one of valid physical IDs backing this logical multi-camera.
If this key is not available for a logical multi-camera, the camera device implementation
may still switch between different active physical cameras based on use case, but the
current active physical camera information won't be available to the application.
Staring from HIDL ICameraDevice version 3.5, the tag must be available in the capture
result metadata to indicate current active physical camera ID.
OFF
No distortion correction is applied.
FAST Lens distortion correction is applied without reducing frame rate
relative to sensor output. It may be the same as OFF if distortion correction would
reduce frame rate relative to sensor.
HIGH_QUALITY High-quality distortion correction is applied, at the cost of
possibly reduced frame rate relative to sensor output.
Mode of operation for the lens distortion correction block.
android.distortionCorrection.availableModes
The lens distortion correction block attempts to improve image quality by fixing
radial, tangential, or other geometric aberrations in the camera device's optics. If
available, the android.lens.distortion field documents the lens's distortion parameters.
OFF means no distortion correction is done.
FAST/HIGH_QUALITY both mean camera device determined distortion correction will be
applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality
correction algorithms, even if it slows down capture rate. FAST means the camera device
will not slow down capture rate when applying correction. FAST may be the same as OFF if
any correction at all would slow down capture rate. Every output stream will have a
similar amount of enhancement applied.
The correction only applies to processed outputs such as YUV, Y8, JPEG, or DEPTH16; it is
not applied to any RAW output.
This control will be on by default on devices that support this control. Applications
disabling distortion correction need to pay extra attention with the coordinate system of
metering regions, crop region, and face rectangles. When distortion correction is OFF,
metadata coordinates follow the coordinate system of
android.sensor.info.preCorrectionActiveArraySize. When distortion is not OFF, metadata
coordinates follow the coordinate system of android.sensor.info.activeArraySize. The
camera device will map these metadata fields to match the corrected image produced by the
camera device, for both capture requests and results. However, this mapping is not very
precise, since rectangles do not generally map to rectangles when corrected. Only linear
scaling between the active array and precorrection active array coordinates is
performed. Applications that require precise correction of metadata need to undo that
linear scaling, and apply a more complete correction that takes into the account the app's
own requirements.
The full list of metadata that is affected in this way by distortion correction is:
* android.control.afRegions
* android.control.aeRegions
* android.control.awbRegions
* android.scaler.cropRegion
* android.statistics.faces
n
List of distortion correction modes for android.distortionCorrection.mode that are
supported by this camera device.
Any value listed in android.distortionCorrection.mode
No device is required to support this API; such devices will always list only 'OFF'.
All devices that support this API will list both FAST and HIGH_QUALITY.
HAL must support both FAST and HIGH_QUALITY if distortion correction is available
on the camera device, but the underlying implementation can be the same for both modes.
That is, if the highest quality implementation on the camera device does not slow down
capture rate, then FAST and HIGH_QUALITY will generate the same output.
FALSE
TRUE
Whether this camera device can support identical set of stream combinations
involving HEIC image format, compared to the
{@link android.hardware.camera2.CameraDevice#createCaptureSession table of combinations}
involving JPEG image format required for the device's hardware level and capabilities.
All the static, control and dynamic metadata tags related to JPEG apply to HEIC formats
as well. For example, the same android.jpeg.orientation and android.jpeg.quality are
used to control the orientation and quality of the HEIC image. Configuring JPEG and
HEIC streams at the same time is not supported.
If a camera device supports HEIC format (ISO/IEC 23008-12), not only does it
support the existing mandatory stream
{@link android.hardware.camera2.CameraDevice#createCaptureSession combinations}
required for the device's hardware level and capabilities, it also supports swapping
each JPEG stream with HEIC stream in all guaranteed combinations.
For every HEIC stream configured by the application, the camera framework sets up 2
internal streams with camera HAL:
* A YUV_420_888 or IMPLEMENTATION_DEFINED HAL stream as input to HEIC or HEVC encoder.
* A BLOB stream with JPEG_APPS_SEGMENTS dataspace to extract application markers
including EXIF and thumbnail to be saved in HEIF container.
A camera device can output HEIC format to the application if and only if:
* The system contains a HEIC or HEVC encoder with constant quality mode, and
* This tag is set to TRUE, meaning that camera HAL supports replacing JPEG streams in
all mandatory stream combinations with a [YUV_420_888/IMPLEMENTATION_DEFINED stream +
JPEG_APPS_SEGMENT BLOB stream] combo.
As an example, if the camera device's hardware level is LIMITED, and it supports HEIC,
in addition to the required stream combinations, HAL must support below stream
combinations as well:
* IMPLEMENTATION_DEFINED/YUV_420_888 MAXIMUM + JPEG_SEGMENTS_BLOB,
* PRIV PREVIEW + IMPLEMENTATION_DEFINED/YUV_420_888 MAXIMUM + JPEG_SEGMENTS_BLOB,
* YUV PREVIEW + IMPLEMENTATION_DEFINED/YUV_420_888 MAXIMUM + JPEG_SEGMENTS_BLOB,
* PRIV PREVIEW + YUV PREVIEW + IMPLEMENTATION_DEFINED/YUV_420_888 MAXIMUM +
JPEG_SEGMENTS_BLOB
The selection logic between YUV_420_888 and IMPLEMENTATION_DEFINED for HAL internal
stream is as follows:
if (HEIC encoder exists and supports the size) {
use IMPLEMENTATION_DEFINED with GRALLOC_USAGE_HW_IMAGE_ENCODER usage flag;
} else {
// HVC encoder exists
if (size is less than framework predefined tile size) {
use IMPLEMENTATINO_DEFINED with GRALLOC_USAGE_HW_VIDEO_ENCODER usage flag;
} else {
use YUV_420_888;
}
}
The maximum number of Jpeg APP segments supported by the camera HAL device.
The camera framework will use this value to derive the size of the BLOB buffer with
JPEG_APP_SEGMENTS dataspace, with each APP segment occupying at most 64K bytes. If the
value of this tag is n, the size of the framework allocated buffer will be:
n * (2 + 0xFFFF) + sizeof(struct CameraBlob)
where 2 is number of bytes for APP marker, 0xFFFF is the maximum size per APP segment
(including segment size).
The value of this tag must be at least 1, and APP1 marker (0xFFE1) segment must be the
first segment stored in the JPEG_APPS_SEGMENTS BLOB buffer. APP1 segment stores EXIF and
thumbnail.
Since media encoder embeds the orientation in the metadata of the output image, to be
consistent between main image and thumbnail, camera HAL must not rotate the thumbnail
image data based on android.jpeg.orientation. The framework will write the orientation
into EXIF and HEIC container.
APP1 segment is followed immediately by one or multiple APP2 segments, and APPn
segments. After the HAL fills and returns the JPEG_APP_SEGMENTS buffer, the camera
framework modifies the APP1 segment by filling in the EXIF tags that are related to
main image bitstream and the tags that can be derived from capture result metadata,
before saving them into the HEIC container.
The value of this tag must not be more than 16.
n
4
OUTPUT
INPUT
The available HEIC (ISO/IEC 23008-12) stream
configurations that this camera device supports
(i.e. format, width, height, output/input stream).
The configurations are listed as `(format, width, height, input?)` tuples.
If the camera device supports HEIC image format, it will support identical set of stream
combinations involving HEIC image format, compared to the combinations involving JPEG
image format as required by the device's hardware level and capabilities.
All the static, control, and dynamic metadata tags related to JPEG apply to HEIC formats.
Configuring JPEG and HEIC streams at the same time is not supported.
All the configuration tuples `(format, width, height, input?)` will contain
AIMAGE_FORMAT_HEIC format as OUTPUT only.
These are output stream configurations for use with dataSpace HAL_DATASPACE_HEIF.
Do not set this property directly. It is populated by camera framework and must not be
set by the HAL layer.
4
n
This lists the minimum frame duration for each
format/size combination for HEIC output formats.
(format, width, height, ns) x n
This should correspond to the frame duration when only that
stream is active, with all processing (typically in android.*.mode)
set to either OFF or FAST.
When multiple streams are used in a request, the minimum frame
duration will be max(individual stream min durations).
See android.sensor.frameDuration and
android.scaler.availableStallDurations for more details about
calculating the max frame rate.
Do not set this property directly. It is populated by camera framework and must not be
set by the HAL layer.
4
n
This lists the maximum stall duration for each
output format/size combination for HEIC streams.
(format, width, height, ns) x n
A stall duration is how much extra time would get added
to the normal minimum frame duration for a repeating request
that has streams with non-zero stall.
This functions similarly to
android.scaler.availableStallDurations for HEIC
streams.
All HEIC output stream formats may have a nonzero stall
duration.
Do not set this property directly. It is populated by camera framework and must not be
set by the HAL layer.