Gwenole Beauchesne [Fri, 13 Apr 2012 14:20:35 +0000 (16:20 +0200)]
Add initial JPEG decoding API.
This is work-in-progress API subject to changes. In particular, this
API is only suitable to Baseline profile. Huffman and quantization
tables also need to be refined.
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Ermilov, Dmitry [Thu, 19 Apr 2012 00:50:34 +0000 (08:50 +0800)]
vatrace: trace H.264 encode parameters
Signed-off-by: Ermilov, Dmitry <dmitry.ermilov@intel.com>
Gwenole Beauchesne [Fri, 6 Apr 2012 09:25:34 +0000 (11:25 +0200)]
API: h264 decode: fix the definition of slice_data_bit_offset.
Align with the existing practice in various media stacks available
in Linux today. i.e. codec layer submits slice data that represents
the original bitstream, thus including any emulation prevent bytes.
However, the slice_data_bit_offset represents an offset relative
to the buffer that contains the NAL unit byte and the rest of the
slice with any emulation prevention bytes removed.
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Gwenole Beauchesne [Sun, 8 Apr 2012 05:19:53 +0000 (07:19 +0200)]
Revert "API: h264 decode: introduce num_emulation_prevention_bytes."
This change turns out to be useless based on a survey of all existing
drivers and codec layers implemented so far. It's better to stick to
the behaviour whereby slice_data_bit_offset excludes emulation prevention
bytes count. Let's schedule this change is this is really necessary.
This reverts commit
a870f078984ce829f797e5610b30d39a3921af4f.
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Gwenole Beauchesne [Thu, 1 Mar 2012 13:01:33 +0000 (14:01 +0100)]
API: h264 decode: introduce num_emulation_prevention_bytes.
num_emulation_prevention_bytes represent the number of emulation prevention
bytes available in the slice_header(). This helps VA drivers determine how
many emulation prevention bytes there are so that to program the hardware
appropriately, without having to parse the slice_header() again.
Rationale: it is not always possible to parse the slice_header() again.
For instance, if "protected slices" are used, the slice data is uploaded
into a VAProtectedSliceDataBuffer and it is no longer possible to read
bytes from it afterwards. e.g. it could be in encrypted memory region
or protected memory region only accessible from the video decoder.
In newly written code, this member shall be filled in appropriately along
with emulation_prevention_bytes_flag set to 1.
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Gwenole Beauchesne [Tue, 3 Apr 2012 08:08:59 +0000 (10:08 +0200)]
API: h264 decode: clarify slice_data_bit_offset field.
slice_data_bit_offset represents a bit offset from the NAL Header Unit
byte to the begining of the slice_data(), thus including any emulation
prevention bytes. Baiscally, this represents the number of bits parsed
for slice_header().
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Gwenole Beauchesne [Tue, 3 Apr 2012 08:36:22 +0000 (10:36 +0200)]
API: vpp: remove extraneous flags member.
VAProcPipelineCaps contained an extra flags member which existed before
its split into pipeline_flags and filter_flags.
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Gwenole Beauchesne [Tue, 3 Apr 2012 14:42:23 +0000 (16:42 +0200)]
API: bump version to 0.34.0 (libva 1.2.0).
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Xiang, Haihao [Wed, 29 Feb 2012 01:03:49 +0000 (09:03 +0800)]
test/transcode/mpeg2transcode: set right picture width/height in macroblocks
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
f8be4a5477e58c4a27170f0c87eeedac8de60aef)
Zhou Chang [Thu, 22 Mar 2012 07:41:10 +0000 (15:41 +0800)]
Added high profile support in SPS for avcenc.
Jonathan Bian [Sat, 3 Mar 2012 00:29:41 +0000 (16:29 -0800)]
Added aspect ratio related fields to VAEncSequenceParameterBufferH264
Signed-off-by: Jonathan Bian <jonathan.bian@intel.com>
Austin Yuan [Thu, 1 Mar 2012 06:48:42 +0000 (22:48 -0800)]
VACodedBufferSegment: correct the description
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 06:21:12 +0000 (22:21 -0800)]
va_TraceCreateSurface->va_TraceCreateSurfaces: correct name change
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:59:03 +0000 (21:59 -0800)]
Android.mk: change Android Makefile accordingly
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:45:10 +0000 (21:45 -0800)]
test: use new vaCreateSurfaces interface
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:44:02 +0000 (21:44 -0800)]
VAProcPipelineParameterBufferType: start from 41 in case there are more encode buffer types to be added in future
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:41:40 +0000 (21:41 -0800)]
VA_STATUS_ERROR_HW_BUSY: error code to indicate HW is busy
Example: the harward only supports one way encode, but we
start two encode apps. This error code can return to app
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:37:34 +0000 (21:37 -0800)]
VAEncMiscParameterRateControl: add back some fields and refine the comments
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:34:31 +0000 (21:34 -0800)]
VADisplayAttribDirectSurface: drop it
vaPutSurface may use overlay or GPU alternately. App should
use vaQuerySurfaceStatus to check if the surface can be used
for decoding. If VASurfaceDisplaying is returned, it means
the surface is being rendered by overlay and can't be over-writeen.
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Thu, 1 Mar 2012 05:34:06 +0000 (21:34 -0800)]
VADisplayAttribBLEMode: drop it since no implementation for it
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Austin Yuan [Tue, 28 Feb 2012 00:46:16 +0000 (19:46 -0500)]
VACodedBufferSegment: modify the semantics/description
VACodedBufferSegment may contain more than one NALs, and flag
VA_CODED_BUF_STATUS_SINGLE_NALU will be used to singal the
segment only contain one NAL
Signed-off-by: Austin Yuan <shengquan.yuan@intel.com>
Xiang, Haihao [Thu, 23 Feb 2012 02:50:24 +0000 (10:50 +0800)]
test/transcode: Fix DSO issue
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 22 Feb 2012 01:56:35 +0000 (09:56 +0800)]
test/transcode/mpeg2transcode: use the method of constant qp
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 21 Feb 2012 08:15:41 +0000 (16:15 +0800)]
Fix sample code for VPP
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 16 Feb 2012 06:55:42 +0000 (14:55 +0800)]
install va_enc.h and va_enc_h264.h
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 16 Feb 2012 03:03:20 +0000 (11:03 +0800)]
test: use the new H264 encoding and VPP interfaces
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Hai Lan [Wed, 7 Sep 2011 03:38:54 +0000 (11:38 +0800)]
Add test/transcode/Makefile.am
Signed-off-by: Hai Lan <hai.lan@intel.com>
Xiang, Haihao [Mon, 10 Oct 2011 05:13:59 +0000 (13:13 +0800)]
test: use new interface for all test cases
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Hai Lan [Wed, 7 Sep 2011 08:13:07 +0000 (16:13 +0800)]
test/transcode/mpeg2transcode.cpp:modify the comments
Signed-off-by: Hai Lan <hai.lan@intel.com>
Hai Lan [Wed, 7 Sep 2011 03:27:55 +0000 (11:27 +0800)]
test/transcode: Add a test case for transcoding
It will decode a simple mpeg2 clip(320x240) and scale it to 1280x960,
then it will encode it to a 264 file(test.264).
Signed-off-by: Hai Lan <hai.lan@intel.com>
Zhou Chang [Mon, 6 Feb 2012 15:52:26 +0000 (23:52 +0800)]
test/encode/avcenc: Add HRD support in avcenc demo application.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Xiang, Haihao [Tue, 31 Jan 2012 08:24:30 +0000 (16:24 +0800)]
test/encode/avcenc: support I420/YV12 format
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhou Chang [Thu, 13 Oct 2011 07:10:00 +0000 (15:10 +0800)]
test/encode/avcenc: Add transform 8x8 support in test application.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Xiang, Haihao [Mon, 10 Oct 2011 05:13:59 +0000 (13:13 +0800)]
test: use new interface for all test cases
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Hai Lan [Thu, 25 Aug 2011 07:37:46 +0000 (15:37 +0800)]
test/encode/avcenc: use kbps as fb input
Signed-off-by: Hai Lan <hai.lan@intel.com>
Xiang, Haihao [Fri, 29 Jul 2011 07:34:48 +0000 (15:34 +0800)]
test/encode/avcenc: use new packed header interface
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhou Chang [Fri, 29 Jul 2011 01:15:10 +0000 (09:15 +0800)]
test/encode/avcenc: Wait upload thread exiting before release driver resource.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Zhou Chang [Wed, 27 Jul 2011 07:16:25 +0000 (15:16 +0800)]
test/encode/avcenc: Simplify the IPB pattern file upload
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Zhou Chang [Wed, 27 Jul 2011 06:44:32 +0000 (14:44 +0800)]
test/encode/avcenc: Fixed bug about B frame display order wrong.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Hai Lan [Mon, 25 Jul 2011 07:07:57 +0000 (15:07 +0800)]
avcenc: correct the description for the mode parameter
Signed-off-by: Hai Lan <hai.lan@intel.com>
Xiang, Haihao [Fri, 8 Jul 2011 05:07:23 +0000 (13:07 +0800)]
test/encode/avcenc: close files
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 7 Jul 2011 02:41:20 +0000 (10:41 +0800)]
test/encode/avcenc: re-try until success
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 5 Jul 2011 00:25:39 +0000 (08:25 +0800)]
test/encode/avcenc: upload YUV data in a thread
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 4 Jul 2011 04:43:57 +0000 (12:43 +0800)]
test/encode/avcenc: silence waring about implicit declaration
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 4 Jul 2011 01:43:23 +0000 (09:43 +0800)]
test/encode/avcenc: postpone YUV data uploading
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Fri, 1 Jul 2011 05:39:55 +0000 (13:39 +0800)]
test/encode/avcenc: increase the times of trying after overflow
Also unmap the coded buffer if overflow
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Hai Lan [Fri, 1 Jul 2011 03:35:55 +0000 (11:35 +0800)]
avcenc: use gettimeofday to calculate the FPS
Signed-off-by: Hai Lan <hai.lan@intel.com>
Hai Lan [Fri, 1 Jul 2011 02:32:19 +0000 (10:32 +0800)]
test/encode/avcenc: Fix bug of ignoring QP value
Signed-off-by: Hai Lan <hai.lan@intel.com>
Hai Lan [Thu, 30 Jun 2011 08:39:33 +0000 (16:39 +0800)]
test/encode/avcenc: Fix a QP error in avcenc
Signed-off-by: Hai Lan <hai.lan@intel.com>
Hai Lan [Thu, 30 Jun 2011 08:04:08 +0000 (16:04 +0800)]
test/encode/avcenc: Add a parameter to set I/IP/IPB mode
Signed-off-by: Hai Lan <hai.lan@intel.com
Zhou Chang [Thu, 30 Jun 2011 08:15:52 +0000 (16:15 +0800)]
test/encode/avcenc: Support continue B frames in test code.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Xiang, Haihao [Thu, 30 Jun 2011 07:44:23 +0000 (15:44 +0800)]
test/encode/avcenc: use generic ExtType name in test case
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 30 Jun 2011 02:41:25 +0000 (10:41 +0800)]
test/encode/avcenc: send packed PPS/SPS header to driver
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhou Chang [Thu, 30 Jun 2011 02:07:24 +0000 (10:07 +0800)]
test/encode/avcenc: Modify default direct mode.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Xiang, Haihao [Wed, 29 Jun 2011 12:54:46 +0000 (20:54 +0800)]
test/encode/avcenc: use a small size for coded buffer.
Must check the status of coded buffer, if overflow, enlarge the size
and try again.
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 29 Jun 2011 05:33:05 +0000 (13:33 +0800)]
test/encode/avcenc: directly uses coded buffer size
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 29 Jun 2011 05:02:05 +0000 (13:02 +0800)]
test/encode/avcenc: directly save slice data
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 29 Jun 2011 01:30:35 +0000 (09:30 +0800)]
test/encode/avcenc: use trail data to calculate slice length
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 28 Jun 2011 07:31:00 +0000 (15:31 +0800)]
test/encode/avcenc: don't generate slice header anymore
It is generated by driver
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 28 Jun 2011 02:04:09 +0000 (10:04 +0800)]
test/encode/avcenc: pass all parameters to driver
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 20 Jun 2011 07:14:20 +0000 (15:14 +0800)]
test/encode/avcenc: use new buffer type to pass parameters
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 20 Jun 2011 06:58:23 +0000 (14:58 +0800)]
test/encode/avcenc: indentation fix
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhou Chang [Tue, 14 Jun 2011 03:01:47 +0000 (11:01 +0800)]
test/encode/avcenc: Add bit rate control/const qp mode in test code.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Zhou Chang [Wed, 8 Jun 2011 07:24:04 +0000 (15:24 +0800)]
test/encode/avcenc: fix reading raw YUV file order bug.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Zhou Chang [Tue, 7 Jun 2011 02:41:38 +0000 (10:41 +0800)]
test/encode/avcenc: bug fixed: B frame stream can't be decoded by JM.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Zhou Chang [Thu, 2 Jun 2011 08:47:09 +0000 (16:47 +0800)]
test/encode/avcenc: Support B frame in MFC and VME.
Signed-off-by: Zhou Chang <chang.zhou@intel.com>
Xiang, Haihao [Thu, 16 Feb 2012 01:53:43 +0000 (09:53 +0800)]
Add VAEncMiscParameterHRD data structure to pass HRD parameters to driver
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Gwenole Beauchesne [Wed, 1 Feb 2012 23:13:48 +0000 (00:13 +0100)]
va: fix vaErrorStr() to support new error codes.
Xiang, Haihao [Mon, 1 Aug 2011 00:41:37 +0000 (08:41 +0800)]
vainfo: silence compiler warning
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 5 Jul 2011 07:46:55 +0000 (15:46 +0800)]
vainfo: update vainfo for Video Proc Pipeline
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Gwenole Beauchesne [Fri, 18 Nov 2011 12:49:32 +0000 (12:49 +0000)]
vpp: link modules from the main page.
Gwenole Beauchesne [Mon, 2 Jan 2012 16:44:53 +0000 (17:44 +0100)]
Allow 0.32.x drivers.
Gwenole Beauchesne [Mon, 2 Jan 2012 16:31:17 +0000 (17:31 +0100)]
API: bump minor version.
Gwenole Beauchesne [Mon, 2 Jan 2012 16:28:45 +0000 (17:28 +0100)]
API: add compat glue for older vaCreateSurfaces().
Gwenole Beauchesne [Sun, 18 Dec 2011 22:51:46 +0000 (23:51 +0100)]
tests: cope with new vaCreateSurfaces() API.
Gwenole Beauchesne [Sun, 18 Dec 2011 22:50:25 +0000 (23:50 +0100)]
API: add surface attributes.
Gwenole Beauchesne [Sun, 18 Dec 2011 17:44:49 +0000 (18:44 +0100)]
vpp: add video processing hooks.
Gwenole Beauchesne [Sun, 18 Dec 2011 17:04:24 +0000 (18:04 +0100)]
vpp: fix comments for *Count enumerations.
Gwenole Beauchesne [Sun, 18 Dec 2011 17:00:54 +0000 (18:00 +0100)]
vpp: allow input/output color primaries as requested.
XXX: this is still a complicated usage model. The color primary could
be a surface atttribute instead.
Gwenole Beauchesne [Sun, 18 Dec 2011 16:39:45 +0000 (17:39 +0100)]
vpp: move reference frames to pipeline parameter buffer.
Gwenole Beauchesne [Sun, 18 Dec 2011 16:24:50 +0000 (17:24 +0100)]
vpp: note that application of subpictures is now optional.
Gwenole Beauchesne [Sun, 18 Dec 2011 16:22:58 +0000 (17:22 +0100)]
vpp: add pipeline flags.
Gwenole Beauchesne [Sun, 18 Dec 2011 16:12:09 +0000 (17:12 +0100)]
vpp: explain filter flags more.
Gwenole Beauchesne [Sun, 18 Dec 2011 15:22:06 +0000 (16:22 +0100)]
vpp: add auto-saturation, auto-brightness, auto-contrast attributes.
Gwenole Beauchesne [Sun, 18 Dec 2011 15:06:32 +0000 (16:06 +0100)]
vpp: clarify background color semantics and driver behaviour.
Gwenole Beauchesne [Sun, 18 Dec 2011 14:56:30 +0000 (15:56 +0100)]
vpp: keep VAProcFilterParameterBuffer value type as float.
Gwenole Beauchesne [Sun, 18 Dec 2011 14:56:05 +0000 (15:56 +0100)]
vpp: fix vaQueryVideoProcPipelineCaps() semantics.
Gwenole Beauchesne [Mon, 28 Nov 2011 13:46:32 +0000 (13:46 +0000)]
vpp: mention that source surface also covers its associated subpictures.
Gwenole Beauchesne [Mon, 28 Nov 2011 13:38:33 +0000 (13:38 +0000)]
vpp: improve documentation for VAProcPipelineParameterBuffer.
Gwenole Beauchesne [Mon, 28 Nov 2011 12:36:10 +0000 (12:36 +0000)]
vpp: fix vaQueryVideoProcFilterCaps() prototype.
Gwenole Beauchesne [Thu, 24 Nov 2011 14:24:36 +0000 (14:24 +0000)]
vpp: remove obsolete sample code.
Gwenole Beauchesne [Thu, 24 Nov 2011 14:23:46 +0000 (14:23 +0000)]
vpp: fix VAProcPipelineParameterBuffer typedef.
Gwenole Beauchesne [Thu, 17 Nov 2011 19:08:15 +0000 (19:08 +0000)]
API: add video processing interfaces.
Gwenole Beauchesne [Sun, 18 Dec 2011 18:13:43 +0000 (19:13 +0100)]
API: encode: document VACodedBufferSegment.
Make it clear that there shall be one VACodedBufferSegment per NAL unit.
It's up to the driver to calculate the various offsets to the NAL units.
Gwenole Beauchesne [Sun, 18 Dec 2011 18:03:25 +0000 (19:03 +0100)]
API: h264 encode: add VA_RC_VBR_CONSTRAINED.
Gwenole Beauchesne [Sun, 18 Dec 2011 18:00:45 +0000 (19:00 +0100)]
API: h264 encode: fix VAConfigAttribEncMaxRefFrames semantics.
Gwenole Beauchesne [Thu, 24 Nov 2011 17:04:10 +0000 (17:04 +0000)]
API: h264 encode: add more explicit packed header types for H.264.
Gwenole Beauchesne [Thu, 24 Nov 2011 16:50:42 +0000 (16:50 +0000)]
API: h264 encode: add packed SEI header.
Gwenole Beauchesne [Thu, 24 Nov 2011 14:05:59 +0000 (14:05 +0000)]
API: h264 encode: make bitstream writers more consistent.
Avoid downcasts to the generic bitstream writer, libva will handle
(and check this) internally.
Gwenole Beauchesne [Thu, 24 Nov 2011 13:57:41 +0000 (13:57 +0000)]
API: h264 encode: add macroblock-info config attribute.
projects / profile / ivi / libva.git / log