| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
| |
This is not needed for today's sensors as chromatic aberration is not an issue.
This saves ~160 mS in the shot to jpeg time
Change-Id: I24633a448b6eed615503bf306c3b3c9f62027fd3
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
| |
For real faces, it is seen that the h/w passes a score >=80
For false faces, we seem to get even a score of 70 sometimes.
In order to avoid any issue at application level, we filter
<=70 score.
Change-Id: Ia4855fd79559e6eb8398d80479cbafbd343321f7
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a NULL window is passed in setPreviewDisplay, we try to do cancelBuffer
to return the buffers to the ANativeWindow if there is a valid reference
to DisplayAdapter. However, the CameraService already sets the ANativeWindow
reference to NULL before calling setPreviewWindow method of the Camera HAL
so the cancel_buffer crashes. So, we do not call cancel_buffer as the
preview window is destroyed anyway by Camera Service.
Change-Id: Ie07f5fd3ca5bb5a31cb25a174ee1acc2988628af
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
| |
Signed-off-by: Solaiyappan Saravanan <saravanan.s@ti.com>
Change-Id: Ie3ed733fb163ac686c9d42766923ccc7e3715a5e
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Previously during CAF any calls for doing AF were
immediately returning status. The patch will check if
CAF has succeeded in locking and if this is not the case
it will temporarily switch to 'auto' triggering a full
AF sweep. CAF is again restored afterwards.
- Additionally this patch adds support for canceling AF
during 'PREVIEW_STATE' state, which is needed in order
to release the 3A lock after AF
Change-Id: Ie8c9b77fb134e63ed668a22453c1672b34488d94
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
| |
1. Face prioritized 3A is enabled for AF, AE (for AE it is disabled)
2. White balance modes, exposure are allowed to be changed when face detection is ON
3. Focus metering area is ignored when face detection is ON
4. CancelAF is fixed to not retain the focus mode for CAF and infinity modes
Change-Id: I43650069d4de95815381974072f5b9afabf4fc46
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
- The recording hint implementation was bypassing
the logic for restoring the original correct
CameraParameters, which is needed for passing the
test.
Signed-off-by: Emilian Peev <epeev@mm-sol.com>
Change-Id: I8c119153dfedab5511955bff1d4c1193014ed720
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- If the client releases the camera when the
ANativeDisplayAdapter is paused, then a segfault
is possible. This is due to the deferred
deinitialization of ANativeDisplaAdapter in it's
destructor. By this time 'mPreviewWindow' in
'CameraHardwareInterface' is already NULL and
all further calls involving it will result in a
segfault.
- Contains a fix for several CTS tests including
#testJpegThumbnailSize and #testJpegExif.
Change-Id: I030e8fdccdacc6d3bf08550da562eac90b50adc3
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Allows stacking of different algorithm priorities.
- Resolves an issue with AF not functioning after
one iteration.
- Contains a couple of fixes related to WB and
Exposure priority modes.
Signed-off-by: Emilian Peev <epeev@mm-sol.com>
Change-Id: I9c2848e776c727cd9cc3691221621f895dc5507b
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Enable face priority 3A when starting face detection
and reset to 3A parameters from CameraParameters when
stopping face detection. Ignore 3A parameters set by
application if face detection is started.
Signed-off-by: Tyler Luu <tluu@ti.com>
Change-Id: I54ded2b3612f3191f6f4c3a6151ea6013b1d06a0
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- It's possible for the camera client to "abandon" the
preview surface at any time. In this case the SurfaceTexture
will start returning ENODEV error codes on every call.
ANativeWindowDisplayAdapter should handle this case and
not try to call the SurfaceTexture any more. If this is
not the case, then there is a chance of a segfault in
MediaServer.
Signed-off-by: Emilian Peev <epeev@mm-sol.com>
Change-Id: Ic529cffae5315cbe6316e583b399d3bd7a44b75c
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
| |
In some cases mFaceData is NOT null and mFaceData->faces is NULL.
In such cases we end up not freeing mFaceData. This patch takes
care of this case and frees mFaceData in such cases.
Change-Id: Ia73161064bec6f30f947f3ab596e09354e9d4b6d
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
| |
Change-Id: I1320bb9dbdf92ad1f194b32ff78aef7088d326ff
Original-author: Akwasi Boateng <akwasi.boateng@ti.com>
Signed-off-by: Akwasi Boateng <akwasi.boateng@ti.com>
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch adds HAL support for FOCUS_MODE_CONTINUOUS_PICTURE
which is needed for enabling CAF in still image preview.
Although the HAL behavior is same for both VIDEO and STILL
CAF, by setting EFocusControlAuto for ducati, the ducati
distinguishes STILL and VIDEO CAF internally based on
Operating mode (pipeline that is selected - video or zsl)
Change-Id: I808d41620caacca94dd65985c467f41417dee6d8
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
| |
Change-Id: Iacf80f7485e78d20000df27bca9e57caafcea111
Signed-off-by: Iliyan Malchev <malchev@google.com>
|
|
|
|
|
|
|
|
|
| |
With the SW usage flags gralloc is enabling caching for TILER buffers
and using wrong memsets in the code which is causing DMM errors and results in
reboots in es1.1 silicon.
Change-Id: I819b40fee0521bafadff1d72c74a7b30d1de9a8e
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
| |
Change-Id: Ia63104950fdfbaf41254187dd0ee516fde8eb7d2
Signed-off-by: Iliyan Malchev <malchev@google.com>
|
|
|
|
|
|
|
|
| |
FD callback behavior is updated in latest javadoc
patch "Improve camera face detection javadoc."
Change-Id: I963628ae55eec62083c25d34bbb380ae6673e431
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
|
|
|
|
|
| |
When sensor is in landscape orientation, we should not be
accounting for the mirroring in the sensor orientation
calculations.
bug:5183013
Change-Id: I3bc854c79d210622d99edaf0718c9ba0ca40ceee
|
|
|
|
| |
This reverts commit 2f394067996c47b8c02a335af7403c34908bff07.
|
|
|
|
| |
This reverts commit ba9c1504171d4455f0616c317a1bc42e38f1feb9.
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Sanity checks should be performed on the individual
areas for the test to pass.
- On invalid configuration CameraParameters should
revert to the last known good configuration.
Change-Id: I9026f287e884835a98c835e8f949ad4441c030bd
Original-author: Emilian Peev <epeev@mm-sol.com>
Signed-off-by: Emilian Peev <epeev@mm-sol.com>
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
|
|
|
|
|
| |
- Should resolve an issue seen with CTS test
#testLockUnlock.
Change-Id: I99e7f36e3468162b0f4056c07c6f9307384354df
Original-author: Emilian Peev <epeev@mm-sol.com>
Signed-off-by: Emilian Peev <epeev@mm-sol.com>
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Framerate ranges were not ordered as per CTS requirement:
( maxFps1 < maxFps2 ) || (maxFps1 == maxFps2 && minFps1 < minFps2)
- This modification also gives higher priority to framerate
ranges over the deprecated constant fps.
Change-Id: Idee8b7ae9c1a3ce1d947bc5067c2c787f30e4d39
Original-author: Emilian Peev <epeev@mm-sol.com>
Signed-off-by: Emilian Peev <epeev@mm-sol.com>
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Returns error for the invalid parameters in
setparameters().
Fixes a cts test failure(#testInvalidParameters).
Change-Id: I11ae111404b64641ccc0ac89f5abe854acb18c35
Original-author: Solaiyappan Saravanan <saravanan.s@ti.com>
Signed-off-by: Solaiyappan Saravanan <saravanan.s@ti.com>
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Need to translate FD results when phone is 180
degree oriented with respect to the sensor. Ducati
outputs results differently then what Android
expects in this orientation. See code comments.
Change-Id: Ia17b3f03ee37b5ac988993024bc5c989cff96df2
Signed-off-by: Tyler Luu <tluu@ti.com>
bug:5140593
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Recoding Hint is set now by Camcorder App, in order to allow CameraHal
to start preview is Recording specific mode (i.e. Video-Mode).
This patch adds support to handle Recoding Hint in order to set/reset
Recording specific parameters.
This patch also takes care of following contraints set by
Recording Hint support in framework.
1) Camera App can set/reset Recording Hint while preview is active.
2) Camera App can start recording even when the Recording Hint is false.
Change-Id: I6332c316ddc3ddda7f8c9c5739f81e6317c7deb0
Signed-Off-By: Kalpesh Patil <kalpeshpatil@ti.com>
|
|
|
|
|
|
|
|
| |
Enable ldc-nsf for still image preview to
attain good quality.
Change-Id: I44e73af79ba983c938aae4ef01c65fec39478f3a
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
| |
This patch enables of VSTAB & VNF for Front Camera.
The issue with VSTAB & VNF for Front Camera for 720p resolution
has been resolved on Ducati side.
So patch-I28a81696 is being reverted by this patch.
Change-Id: Ie723b2e2a59d0dec7aefd5c9362e24b0dd2f5562
Signed-Off-By: Kalpesh Patil <kalpeshpatil@ti.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes an issue with secondary camera video record - the capture
mode was not being set after switching from image mode because
the variable mWaitToSetConfig was not reset as the camera was
switched within 100 frames.
Should be reverted when the below patch is reverted
2f394067996c (TEMP) Avoid SetConfig when starting front sensor
Change-Id: I8cff990df95af4f4871f434e21ddd4c1d2d8e919
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
| |
The zoom value being passed up with stopSmoothZoom was 0
instead of 1. CTS expects the value of the zoom callback
to atleast start from 1. This patch fixes this.
Change-Id: I2a043af12d15e83012d1e5b633209ccf194e71ea
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
| |
focal length, horizontal and vertical angles
Addresses failures of following tests:
Change-Id: I36107b8c5cffe843ce7d167a607b9f9269f3ed26
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
| |
There could be a potential issue if a CameraHal instance
tries to acquire an OMX Camera handle while another instance
is freeing it.
Change-Id: I33e4c679efc11561ceca3aee217952939bfad6b7
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixing a few robustness issues with smooth zoom.
1) Notify application about smooth zoom success only after
setting the state internally. If we notify application
first, it can signal another smooth zoom start breaking
our state.
2) If we notify in goal reached, we don't have return
zoom status again. Double notification can lead to crash
Change-Id: Idda594957822373cb3fd73f1a00459dd9900d955
Signed-off-by: Tyler Luu <tluu@ti.com>
|
|
|
|
|
| |
Change-Id: I12f1037be3d890b50c5413925d3084a2cf5256aa
Signed-off-by: Sundar Raman <sunds@ti.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Delay SetConfig calls for the first 99 frames when
starting front sensor. Front sensor is currently freezing
if any SetConfig is called within those first frames.
This patch needs to be removed once Ducati side issue
is fixed.
Change-Id: Ie50dab38724077443719a1061c9b038f59c6b545
|
|
Change-Id: I8f7a7eeece0e516efa486b77e9d97805c0e65d3e
Signed-off-by: Iliyan Malchev <malchev@google.com>
|