[Libav-user] Problems with ffvhuff and huvvyuv
bruce at spearmorgan.com
Wed Oct 16 18:16:41 CEST 2013
On Oct 16, 2013, at 8:56 AM, Bruce Wheaton <bruce at spearmorgan.com> wrote:
>> On Oct 16, 2013, at 8:46, Paul B Mahol <onemda at gmail.com> wrote:
>>> The only workaround is to use intra-frame only compression, so all frames
>>> are independent. Some codecs you listed always work that way.
>> ffvhuff is intra only.
> I don't know that codec. I just knew it was lossless. So I guess (unless there's a bug) the rule has an exception. Maybe the 'adaptive tables' that google says it has adapt every 24 or so frames. Don't have the source in front of me right now.
> But it's an FOSS project - anyone look at the code and see what's going on.
I can't pick it out of the code, and the codec info says it is intra-frame only... but this paper seems to indicate that ffvhuff isn't always intra-frame only:
"FFv1  was also developed by the FFmpeg project and is derived from HuffYUV. The main difference is that the intra-frame coding model is limited to the median model as described above"
From COMPARISON OF LOSSLESS VIDEO CODECS FOR CROWD-BASED QUALITY ASSESSMENT ON TABLETS
Christian Keimel, Christopher Pangerl and Klaus Diepold
So - as I mentioned - videos that aren't encoded as intra-frame often have to decode a number of frames in order to get to your desired frame. As far as I know, the only way ffmpeg deals with that is by decoding all the frames from a known safe search point up until your desired frame.
>>>> Does this have anything to do with cached frames, or is that another
>>>> Libav-user mailing list
>>>> Libav-user at ffmpeg.org
>> Libav-user mailing list
>> Libav-user at ffmpeg.org
> Libav-user mailing list
> Libav-user at ffmpeg.org
More information about the Libav-user