Environment: -------------- Intel driver: commit 84bc0b4cd257404a0fd47186b3459df6d29abb5b Author: Li,Xiaowei <xiaowei.a.li@intel.com> Date: Tue Sep 25 13:09:13 2012 +0800 Libva: commit 11ef676d6722b5d203b2eff2639de515510c5673 Author: Gwenole Beauchesne <gwenole.beauchesne@intel.com> Date: Thu Oct 4 15:28:07 2012 +0200 Bug Info: -------------- de-interlace doesn’t work well for h264 bitstream. It seems that gst-vaapi h264 decoder hasn't consider filed/frame difference yet. Reproduce steps: ---------------- 1. xinit & 2. gst-launch-0.10 filesrc location=/root/TS720x480i30_4x3_h264cav_3mbs_daisy-0xe0-950frames.264 ! h264parse ! vaapidecode ! vaapipostproc deinterlace=1 deinterlace-method=1 hue=150.0 brightness=50.0 contrast=5.0 saturation=5.0 auto-brightness=1 auto-contrast=1 auto-saturation=1 ! video/x-surface,type=vaapi,format=\(fourcc\)YV12 ! vaapisink
Changed summary to "add support for H.264 interlaced streams" and set the category as "enhancement". Assigned, work-in-progress locally.
Libva (staging) c59c512df262cdc44fb6e9101621f246eda176df Inter-driver (staging) 3bd6f3b3611dd9a39a3075f82cf938fa08e22884 Gst-vaapi (qa) e7d517323fd98608918de1211dc42f795b48765f Platfrom: IVB Still not support h264 bitstream.
Support for H.264 interlaced streams is now available in git master branch. Bug will be closed unless you find further issues.
Restest the bug in gstream-vaapi master branch (cb0da4ed7acbb51f2d5dba34c0b37b2c5ad8bff4): 1. when using gst-launch-0.10 filesrc location=/root/TS720x480i30_4x3_h264cav_3mbs_daisy-0xe0-950frames.264 ! h264parse ! vaapidecode ! vaapisink to play the interlace file, the video is not smootly and a warning always shown: WARNING: from element /GstPipeline:pipeline0/GstVaapiSink:vaapisink0: A lot of buffers are being dropped. Additional debug info: gstbasesink.c(2875): gst_base_sink_is_too_late (): /GstPipeline:pipeline0/GstVaapiSink:vaapisink0: There may be a timestamping problem, or this computer is too slow. 2. VPP doesn't work. 3. Still not support some interlace file. pls. refer to the #bug 56652
I think it should be supported now, right?
it has been done by gwenole. two fields (top & bottom) are decoded to same vaSurface, it also means BOB deinterlace is done intrinsic.
(In reply to comment #6) > it has been done by gwenole. > two fields (top & bottom) are decoded to same vaSurface, it also means BOB > deinterlace is done intrinsic. Well, it depends. This is the case because I still haven't come by videos with specific SEI messages that report the temporal order of either field. This means that you could have a whole surface decoded made up of two fields but sometimes, with some explicit SEI messages, you can specify which field to display first. i.e. the surface does not necessarily represent the same scene: one field could represent the scene at T, and the other field could represent the scene at T+1. I would be interested if QA could collect some interlaced bitstreams with PicTiming SEI messages. Thanks.
Still some interlcase bitstream can't be decoded normally. e.g. CAMASL3_Sony_B.jsv CAMACI3_Sony_C.jsv cama3_vtc_b.avc cama2_vtc_b.avc cama1_vtc_c.avc Platform: IVB (master branch)
Halley, please check again.
Issue can't be reproduced. Close it.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.