Recherche avancée

Sur d’autres sites (13561)

• Revision fe5e2a91bb : Auto min and max partition size experiment. Speed feature experiment to set an

  24 juillet 2013, par Paul Wilkins

  Changed Paths :
   Modify /vp9/encoder/vp9_encodeframe.c

  
   Modify /vp9/encoder/vp9_onyx_if.c

  
   Modify /vp9/encoder/vp9_onyx_int.h

  
  
  Auto min and max partition size experiment.

  Speed feature experiment to set an upper and lower
  partition size limit based on what has been seen
  in spatial neighbors.

  This seems to gives quite reasonable speed gains in local
  (10-15%) and when used with speed 0 the losses are small
  (0.25% derf, 0.35% stdhd). However, for now I am only
  enabling it on speed 1 as there may be clashes with the existing
  temporal partition selection in speed 2.

  Using a tighter min / max around the range derived from the
  neighbors increases speed further but at the cost of a
  bigger quality loss. However, I think this spatial method could
  be combined with data from either the last frame or a variance
  method (or both) to refine the range of minimum and maximum
  partition size. I.e. consider the min and max from spatial and
  temporal neighbors and the variance recommendation.

  Change-Id : I1b96bf8b84368d6aad0c7aa600fe141b4f07435f

• Encode h264 to match existing stream

  6 juin 2013, par Diego Sánchez

  In short : I have to encode a tiny amount of video frames and stitch them in front of a much bigger h.264 stream without reencoding said stream.

  The details : I receive a multi GB transport stream containing a h.264 es and an audio es. Currently the h.264 streams are always generated using x264, and I can assume this will be the case in the future. Now I have to prepend some video frames to this stream, but am not allowed to decode/encode the whole stream ; which leaves me with the only option to find out the exact parameters I need to pass x264_encoder_open so both streams match.

  Currently what I'm doing is :

  1. Demux the original ts and extract h.264 NAL packets.
  2. When I find the first "user data unregistered" SEI packet, I parse it and find a bunch of x264 parameters.
  3. Use libavcodec to start decoding the video. That gives me the dimensions of the picture and the h264 profile and level in the AVCodecContext structure.
  4. Match all of that as best as I can in the x264_param_t structure.

  I can do some encoding with that, and the encoded video plays correctly up to the join point. When VLC is reaching the stitch point it starts throwing out the following sequence of messages and soon after stops playing :

  [h264 @ 0x7fe36cd75be0] decode_slice_header error
  
  [h264 @ 0x7fe36cd75be0] no frame!
  
  [h264 @ 0x7fe36ccc9080] Width/height changing with threads is not implemented. Update your Libav version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented.

  which clearly shows that my encoded frames do not match the original ones. I've been browsing the source code and can't seem to find a way of doing this. What I have currently (besides not working), involves a lot of guesswork, so even if I could make it work with the handful of example files I have, I would be scared to deploy this in production servers.

  So the obvious question is : Is there a safe, formal way of doing this ?

  Thanks in Advance

• Encode h264 to match existing stream

  6 juin 2013, par Diego Sánchez

  In short : I have to encode a tiny amount of video frames and stitch them in front of a much bigger h.264 stream without reencoding said stream.

  The details : I receive a multi GB transport stream containing a h.264 es and an audio es. Currently the h.264 streams are always generated using x264, and I can assume this will be the case in the future. Now I have to prepend some video frames to this stream, but am not allowed to decode/encode the whole stream ; which leaves me with the only option to find out the exact parameters I need to pass x264_encoder_open so both streams match.

  Currently what I'm doing is :

  1. Demux the original ts and extract h.264 NAL packets.
  2. When I find the first "user data unregistered" SEI packet, I parse it and find a bunch of x264 parameters.
  3. Use libavcodec to start decoding the video. That gives me the dimensions of the picture and the h264 profile and level in the AVCodecContext structure.
  4. Match all of that as best as I can in the x264_param_t structure.

  I can do some encoding with that, and the encoded video plays correctly up to the join point. When VLC is reaching the stitch point it starts throwing out the following sequence of messages and soon after stops playing :

  [h264 @ 0x7fe36cd75be0] decode_slice_header error
  
  [h264 @ 0x7fe36cd75be0] no frame!
  
  [h264 @ 0x7fe36ccc9080] Width/height changing with threads is not implemented. Update your Libav version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented.

  which clearly shows that my encoded frames do not match the original ones. I've been browsing the source code and can't seem to find a way of doing this. What I have currently (besides not working), involves a lot of guesswork, so even if I could make it work with the handful of example files I have, I would be scared to deploy this in production servers.

  So the obvious question is : Is there a safe, formal way of doing this ?

  Thanks in Advance