Recherche avancée

Sur d’autres sites (7347)

• vf_paletteuse : fix color cache lookup for Bayer dithering mode.

  10 janvier 2022, par Rudolf Polzer

  vf_paletteuse : fix color cache lookup for Bayer dithering mode.
  To trigger this bug, use `paletteuse=dither=bayer:bayer_scale=0` ; you will see
  
  that adjacent pixel lines will use the same dither pattern, instead of being
  
  shifted from each other by 32 units (0x20).
  One way to demostrate the bug is :
  $ convert -size 64x256 gradient:black-white -rotate 270 grad.png
  
  $ echo 'P2 2 1 255 0 255' > bw.pnm
  
  $ ffmpeg -i grad.png -filter_complex 'movie=bw.pnm,scale=256x1[bw] ; [0:v][bw]paletteuse=dither=bayer:bayer_scale=0' gradbw.png
  Previously : https://www.rm.cloudns.org/img/uploaded/0bd152c11b9cd99e5945115534b1bdde.png
  
  Now :        https://www.rm.cloudns.org/img/uploaded/89caaa5e36c38bc2c01755b30811f969.png
  This was caused by passing inconsistent color vs (a,r,g,b) parameters to
  
  color_get(), and NBITS being 5 meaning actually hitting the same cache node
  
  does happen in this case, but ONLY if bayer_scale is zero.
  The fix is passing the correct color value to color_get().
  Also added a previous-failing FATE test ; image comparison of the first frame :
  Previously : https://www.rm.cloudns.org/img/uploaded/d0ff9db8d8a7d8a3b8b88bbe92bf5fed.png
  
  Now :        https://www.rm.cloudns.org/img/uploaded/a72389707e719b5cd1c58916a9e79ca8.png
  (on this less synthetic test image, the bug basically causes noise from cache
  
   hits vs misses)
  Tested : FATE passes, which exercises this filter but at the default bayer_scale.
  Reviewed-by : Paul B Mahol <onemda@gmail.com>
  

  • [DH] libavfilter/vf_paletteuse.c
  • [DH] tests/fate/filter-video.mak
  • [DH] tests/ref/fate/filter-paletteuse-bayer0

• avfilter/vf_vectorscope : draw color points names

  7 mars 2016, par Paul B Mahol

  avfilter/vf_vectorscope : draw color points names
  Signed-off-by : Paul B Mahol <onemda@gmail.com>
  

  • [DH] doc/filters.texi
  • [DH] libavfilter/vf_vectorscope.c

• How do I extract color matrix from MP4 an x264 stream in Media Foundation

  23 août 2016, par Jules

  I am playing a video (mp4 containing x264 encoded video stream) with a custom player using media foundation.

  When I convert the YUV information into RGB I need to account for the color matrix and range used at encode time.

  Some of my videos have this information, I can use MediaInfo.exe or FFMPEG to see that it is present.

  However, for such videos if I look at the relevant Media Foundation properties (Extended Color Information) the properties are not present in the files.

  So, somehow I need to find a way to access the information.

  Media Foundation does provide access to MF_MT_MPEG4_SAMPLE_DESCRIPTION and MF_MT_MPEG_SEQUENCE_HEADER for the video stream but I can’t find descriptions of what these contain.

  I noticed that the MF_MT_MPEG_SEQUENCE_HEADER is much longer for the videos with the information present and this (MPEG Headers Quick Reference) seems to suggest headers might contain the information I need.

  I’m looking for Color Range (limited/full), Color Primaries, Transfer Characteristics and Matrix Coefficients (BT.709 etc).

  I’d greatly appreciate any help finding this information from a Media Foundation video stream.

  Thanks

  Jules

  ---

  Update - Sequence Header

  The sequence header appears to be a subset of MPEG4 sample description, though I can’t find anything that indicates what either bits of data actually contains / doesn’t contain specifically.

  The sequence header appears to contain data structured as an MP4 byte stream as described in the H264 Standards Document and includes the VUI (Video Usability Information - Annex E of document) which may then include the colour information I’m interested in.

  Given that it’s a byte stream I need to know where it starts and whether there’s some existing code I could use to decode it.

  In FFMPEG in libavcodec/h264_ps.c there is a function called ff_h264_decode_seq_parameter_set which ends up calling decode_vui_parameters. It seems possible that seq_parameter_set maps to MF_MT_MPEG_SEQUENCE_HEADER and it may be possible to use that code to decode the data.

  If anyone one has any direct experience with decoding this data it would be very useful.

  Thanks again

  ---

  Update - Related posts

  I found this How to decode sprop-parameter-sets in a H264 SDP ? and Possible Locations for Sequence/Picture Parameter Set(s) for H.264 Stream which are fairly helpful.

  The sequence header would appear to be Sequence or picture parameter set (pps) and the parameters I want are the VUI extension subset.

  Plus this post H.264 stream structure gives the high level of how the stream data is structured, and the MF_MT_MPEG_SEQUENCE_HEADER appears to start with a NAL 0x00 0x00 0x01 so I’m guessing it is a NAL containing the PPS.