[FFmpeg-devel] Some questions about ffmpeg h264_qsv decoder

Ivan Uskov ivan.uskov at nablet.com
Wed Sep 16 21:21:08 CEST 2015


Hello Ron,

Wednesday, September 16, 2015, 9:00:02 AM, you wrote:

R> a) qsv decode h264 file found many duplicated frames.
I  have  posted the patch "[PATCH] libavcodec/qsvdec_h2645.c Bug fixed: wrong
ticks_per_frame."  to  this  list, please try to apply it the issue should be
solved.

R> b) A 1920x1080 video transcode to 1280x720, both CPU and GPU used more than 1920x1080 to  1920x1080.
Unfortunately  the  current  implementation of qsv modules of ffmpeg does use
own  standalone  MXF  session. So there are copying from GPU memory to system
memory  and  back  are existing during transcoding. Also here software scaler
uses   when  you  transcode  1920x1080  ==>  1280x720.  We  a  working  under
possibility to create complete pipeline into GPU:  [qsv_dec]==>[qsv_vpp(scaler)]==>[qsv_enc]
But I can not say currently when exactly it will available.



-- 
Best regards,
 Ivan                            mailto:ivan.uskov at nablet.com



More information about the ffmpeg-devel mailing list