[FFmpeg-devel] [RFC]Nvidia VDPAU patches

compn tempn
Mon Dec 1 15:42:25 CET 2008

On Mon, 1 Dec 2008 14:15:34 +0100, Reimar D?ffinger wrote:
>On Mon, Dec 01, 2008 at 12:45:55PM +0100, Andreas Setterlind wrote:
>> including automatic fallback to software
>> decoding (which I understand is something that FFmpeg developers sees
>> as a mandatory requirement before committing any patches?):
>I can not speak for the others but IMO no. As I understood it, Michael
>would reconsider allowing the different PIX_FMTs for different profiles
>if they were necessary to do software fallback, but I disagree with
>that. The only kind of software fallback I'd consider useful is decoding
>in software when the VDPAU decoder reports an error (also useful for
>error concealment, which to my knowledge is horrible in most hardware
>decoders), but I think that will make a huge mess.
>What in my opinion is not acceptable is mandatory reverting to software rendering
>just because a few bits in the file say "oh we use ultra-special
>features with 1 GBit/s bit rate" because in reality it's most likely
>just the encoder being too stupid to set those bits right and the
>hardware could play it just fine (not to mention all those files that
>are just in the higher profile because of some setting that actually
>makes no difference at all to the hardware decoder).

can the vdpau api handle decoding most/all h264 resolutions/bitrate etc?
does the current patch do this? or is it for bluray formats only?

is ffmpeg asking for a generic decoder instead of a per-profile decoder?

sorry for my stupid questions

More information about the ffmpeg-devel mailing list