| a07666d12bfd94b076db1b966d2d10846622309d |
|
26-Oct-2017 |
Junkai Wu <junkai.wu@intel.com> |
Add bowl view mode in stitching.
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| cff638b59385036cbc91873a9965282c8861a306 |
|
25-Sep-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
standard-OCL: switch to CLVideoBuffer path
* CLVideoBuffer inherited from BufferProxy and CLBuffer
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 0995e53b9064a563b343a9bdcd50ecf88bf84bec |
|
23-Aug-2017 |
Wind Yuan <feng.yuan@intel.com> |
module: add stitcher module interface
* module/interface, new feature interface entry.
* add interface/stitcher.h as stitching module interface.
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| d3bee015a4f71561fee16b22beeb3cc9d198d3e8 |
|
30-Jun-2017 |
zongwave <wei.zong@intel.com> |
cl-dvs: add video stabilization using gyroscope
* support Gyro 3-DoF (orientation) stabilization
* test-video-stabilization cmdline:
$ test-video-stabilization --input in.yuv --output out.mp4
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 63dbfb992bae2f9f75f4e94b3f63a3429248a908 |
|
22-Jun-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
cl-stitch: support fisheye lens shading correction
* gst-launch-1.0 cmdline:
$ gst-launch-1.0 filesrc location=input.nv12 \
! videoparse format=nv12 width=4096 height=2048 framerate=24/1 \
! xcamfilter copy-mode=1 enable-stitch=true stitch-scale=local \
stitch-fisheye-map=true stitch-lsc=true stitch-res-mode=4k \
! video/x-raw, foramt=NV12, width=4096, height=2048 \
! queue ! vaapih264enc rate-control=cbr \
! tcpclientsink host="host-ip" port=3000 blocksize=1024000 sync=false
* test-image-stitching cmdline:
$ test-image-stitching --input input.nv12 --output output.mp4 \
--input-w 4096 --input-h 2048 --output-w 4096 --output-h 2048 \
--scale-mode local --enable-fisheyemap --enable-lsc \
--res-mode 4k --framerate 24.0 --save true
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| be505049e0d0cd218324c728b840652ac54bd19f |
|
01-Jun-2017 |
Wind Yuan <feng.yuan@intel.com> |
cl-framework: refine cl kernel handler arguments
* add cl argument class, template and array for different types
for example.
CLArgList args;
args.push_back (new CLMemArgument (cl_image));
args.push_back (new CLArgumentT<int> (1920));
* cl kernel support async mode (disable block), cl arguments
would be released by event hook.
* CLImageKernel arguments can be set by deriving <prepare_arguments> or
directly <set_arguments>. only one can work.
* CLImageHandler, all parameters should be ready in <prepare_parameters>
and released in <execute_done>
* Image input/output buffer can be found via CLImageHandler
get_input_buf ()
get_output_buf ()
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| d502149dd9e27e8b6617ea5ee4659c5d75b3baf6 |
|
23-May-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
cl-stitch: support 4k resolution mode
* tune 4k parameters for video stream
* add debug information to make it easier for qualiy tuning
* support framerate options in test-image-stitching via OpenCV codec
* gst-launch-1.0 cmdline:
$ gst-launch-1.0 filesrc location=input.nv12 \
! videoparse format=nv12 width=4096 height=2048 framerate=24/1 \
! xcamfilter copy-mode=1 enable-stitch=true stitch-scale=local \
stitch-fisheye-map=true stitch-fm-ocl=true stitch-res-mode=4k \
! video/x-raw, foramt=NV12, width=4096, height=2048 \
! queue ! vaapih264enc rate-control=cbr \
! tcpclientsink host="host-ip" port=3000 blocksize=1024000 sync=false
* test-image-stitching cmdline:
$ test-image-stitching --input input.nv12 --output output.mp4 \
--input-w 4096 --input-h 2048 --output-w 4096 --output-h 2048 \
--scale-mode local --enable-fisheyemap --fm-ocl true \
--res-mode 4k --framerate 24.0 --save true
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| c13d2c54fb3648b27aa1f4b67887fbfe23b2272d |
|
12-May-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
cl-stitch: restructure feature match
* extract feature-match as CVFeatureMatch class
* support CPU and GPU/OCL path
* tune parameters to improve quality
* fix crash issue for CPU path when profiling disabled
* gst-launch-1.0 test cmdline:
$ gst-launch-1.0 filesrc location=input.nv12 \
! videoparse format=nv12 width=1920 height=1080 \
! xcamfilter copy-mode=1 enable-stitch=true stitch-scale=local \
stitch-fisheye-map=true stitch-fm-ocl=true \
! video/x-raw, foramt=NV12, width=1920, height=960 \
! queue ! vaapih264enc rate-control=cbr \
! tcpclientsink host="host-ip" port=3000 blocksize=1024000 sync=false
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 01c571d31e2b1c12908ef6519acb9b58008bb296 |
|
18-Apr-2017 |
Wind Yuan <feng.yuan@intel.com> |
stitching: add option for fisheye map
* test cmdline
$ test-image-stitching --input in.nv12 --output out.mp4 --enable-fisheyemap \
--input-w 1920 --input-h 1080 --output-w 1920 --output-h 960 --scale-mode local
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 680275749a2b29042b169e7b35cf4534a638115c |
|
28-Mar-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
cl-stitch: support default stitch info
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 52d45b347e46f480662e5233aec21ec8ca4e6b84 |
|
19-Mar-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
cl-360-stitch: integrate stitching features into module
* xcamfilter support 360-stitch case, gst-launch-1.0 cmdline:
$ gst-launch-1.0 filesrc location=input.nv12 \
! videoparse format=nv12 width=1920 height=1080 \
! xcamfilter copy-mode=1 enable-stitch=true stitch-scale=local \
! video/x-raw, foramt=NV12, width=1920, height=960 \
! queue ! vaapih264enc rate-control=cbr \
! tcpclientsink host="host-ip" port=3000 blocksize=1024000 sync=false
if src capsfilter was not set, default stitch size as follows:
stitch-width = sink-width, stitch-height = sink-width / 2
* test-image-stitching cmdline:
$ test-image-stitching --input input.nv12 --output output.mp4 \
--input-w 1920 --input-h 1080 --output-w 1920 --output-h 960 \
--scale-mode local
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 024a2343d0f365e73fbfbea44ac0a5c2bc8cef40 |
|
27-Feb-2017 |
Yinhang Liu <yinhangx.liu@intel.com> |
log: fix spelling errors/typos
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 2457e457efc09c1139748b41a7b6117f83f50afd |
|
28-Dec-2016 |
zongwave <wei.zong@intel.com> |
cl-3d-denoise: (fix bug) 3D denoise use inappropriate tuning parameter
* fix 3D NR tuning parameter misuse issue
* rename RGB temporal noise reduction tuning parameter to 3D noise reduction
* cmdline:
$ gst-launch-1.0 xcamsrc io-mode=4 imageprocessor=1 analyzer=2 \
pipe-profile=$pipe_index denoise-3d=yuv ! video/x-raw, format=NV12, \
width=1920, height=1080, framerate=25/1 ! queue ! vaapiencode_h264 \
rate-control=cbr ! tcpclientsink host="$host_ip" port=3000 sync=false
Signed-off-by: Zong Wei <wei.zong@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 45cb102eb59497859ba6b5a036dcfa633b6c78bb |
|
15-Nov-2016 |
Wind Yuan <feng.yuan@intel.com> |
cl-framework: fix hang on processor buf manager
* process thread need check handler->is_handler_enabled to decide
next steps.
* change CLImageHandler::set_kernels_enable to enable_handler
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 233fe1a70a00dd094cfa8fe58d5ebdbe91c2e72a |
|
04-Nov-2016 |
zongwave <wei.zong@intel.com> |
cl-warp: add image warp test case in test-cl-image
* test commandline:
$ ./test-cl-image -t warp -f NV12 -g NV12 -i TV_1080P_NV12.yuv
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| 1b10d6f4268b1ca65fe1747c32c9fba1c0343452 |
|
25-Oct-2016 |
zongwave <wei.zong@intel.com> |
cl-warp: enable warp perspective handler
* add warp handler and cl kernel which can work for DVS features
* add structure XCamDVSResult
* test cmdline, need set <buffercount=32 enable-warp=1>
$ gst-launch-1.0 filesrc location=test.nv12 ! videoparse \
width=1920 height=1080 format=nv12 ! xcamfilter copy-mode=1 \
buffercount=32 enable-warp=1 ! queue ! vaapiencode_h264 \
rate-control=cbr ! tcpclientsink host="$ip" port=3000
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| b43f99c0ca903a62838628dc1c06f3d5fb576746 |
|
22-Aug-2016 |
zongwave <wei.zong@intel.com> |
cl-3d-denoise: use sub-group to improve performance
* 2 versions supported
- 1st, use both sub-group and SLM to reach better performance
- 2nd, pure SLM version can work cross platforms and
with good performance.
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| d341dbef88c6c6c3cc1e2a596f8e5c3bcc01e862 |
|
24-Jul-2016 |
Yinhang Liu <yinhangx.liu@intel.com> |
cl-post-processor: support wireframe/scaler/wavelet-nr
* add scaler/wireframe (face detection) testcase in test-pipe-manager
and gst-xcamfilter. Test command line:
# test-pipe-manager --format=NV12 --width=1920 --height=1080 \
--fake-input=input.NV12 --enable-wireframe -p
# gst-launch-1.0 filesrc location=input.NV12 ! videoparse \
width=1920 height=1080 format=nv12 ! xcamfilter copy-mode=1 \
enable-wireframe=true ! queue ! vaapiencode_h264 \
rate-control=cbr ! qtmux ! filesink location=test.mp4
* add wavelet-nr testcase in test-pipe-manager and gst-xcamfilter.
test command line:
# test-pipe-manager --format=NV12 --width=1920 --height=1080 \
--fake-input=input.nv12 --wavelet-mode=X -p
# gst-launch-1.0 filesrc location=input.nv12 ! videoparse \
width=1920 height=1080 format=nv12 ! xcamfilter copy-mode=1 \
wavelet-mode=X ! queue ! vaapiencode_h264 rate-control=cbr \
! qtmux ! filesink location=test.mp4
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| bec89e12eafc649573ed9fa771a2ac1ea4bb54aa |
|
22-Jul-2016 |
zongwave <wei.zong@intel.com> |
cl-3d-denoise: enable 3D denoise CL kernel/handler
* test cmdline:
$ ./test-cl-image -t 3d-denoise -f NV12 -g NV12 \
-i noisy_sequence_nv12.yuv -o 3d_denoise.yuv
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|
| f1a32a2ea07fe016eebaa47a6c1965ac367c5425 |
|
12-Jun-2016 |
Yinhang Liu <yinhangx.liu@intel.com> |
framework: adjust the code skeletons
* move 3alib to plugins/3a
* move smartlib to plugins/smart
* separate modules/isp and modules/ocl from xcore
Signed-off-by: Wind Yuan <feng.yuan@intel.com>
/external/libxcam/modules/ocl/cl_post_image_processor.cpp
|