[FFmpeg-devel] Google Summer of Code participation

[Reimar Döffinger]

On Mon, Mar 30, 2009 at 07:46:31PM +0200, Michael Niedermayer wrote:
> >> If it really is not stored in the frames there are only 3 ways
> >>
> >> 1. change every muxer and demuxer that allows corepng to add&remove a
> >> byte specifying the keyframe and fix pngdec/enc to handle that byte
> >> 2. add a avcodec_decode_video2() that takes a AVPacket and update
> >> ffmpeg.c and ffplay.c (this was planned since a long time and is welcome)
> >> 3. try I & P and compare which is closer to the previous frame
> >> this if course is not 100% but easy
> >>   
> > 1) Well for the decoding it might not make a difference in my eyes, but 
> > when it would come to encoding, this would not comply to the original 
> > CorePNG anymore, would it?
> 
> well, the avi muxer would have to remove the byte again at least, this of
> course is becomes ugly quickly if more (de)muxers use this format.
> 
> The real problem is that these things are designed too quickly and without
> the video/audio expert community being aware, so that when such problems
> are found, its too late to fix.

Well, I think they found them quite quickly themselves, since (AFAICT)
only their VFW decoder supports the P frames, the DirectShow one does
not, and VFW is deprecated since ages...
But could you both please quote a bit more sanely? Hundreds of lines of
text with a one line response is just not at all polite to everyone
else.