[FFmpeg-devel] [PATCH v4 1/2] codec: bitpacked: add decoder

[Damien Riegel]

Hi,

On Fri, Mar 31, 2017 at 07:23:28PM +0100, Rostislav Pehlivanov wrote:
> On 31 March 2017 at 18:11, Kieran Kunhya <kierank at obe.tv> wrote:
> 
> > On Fri, 31 Mar 2017 at 17:57 Rostislav Pehlivanov <atomnuker at gmail.com>
> > wrote:
> >
> > > On 31 March 2017 at 16:36, Damien Riegel <
> > > damien.riegel at savoirfairelinux.com
> > > > wrote:
> > >
> > > > Add a codec capable of decoding some formats of the RFC4175. For now
> > > > it's only capable of handling YCbCr-4:2:2 with 8-bit or 10-bit depth.
> > > >
> > > > For 8-bit it's a simple pass-through, for 10-bit it depacks the stream
> > > > in the AV_PIX_FMT_YUV422P10 pixel format.
> > > >
> > > > Signed-off-by: Damien Riegel <damien.riegel at savoirfairelinux.com>
> > > > ---
> > > > Changes in v4:
> > > >  - use uint64_t for comparison between frame_size and packet_size
> > > >  - add a check to make sure width is even
> > > >  - take into account linesize to compute buffer addresses
> > > >  - use AVERROR_INVALIDDATA instead of custom error codes
> > > >
> > > > Changes in v3:
> > > >  - Codec has been renamed bitpacked (instead of vrawdepay)
> > > >  - A decoding function is now chosen at codec init based on the pixel
> > > >    format
> > > >  - Codec marked as experimental
> > > >
> > > >
> > > Don't mark it as experimental. The experimental flag is only used for
> > > encoders to avoid picking ones which may generate invalid bitstreams
> > > instead of reference implementation external libraries
> > >
> >
> > I asked for it to be marked as experimental, it's not even a real codec, it
> > doesn't handle any of the timestamp difficulties for example during
> > wraparound.
> >
> > Afaik this patch allows for stuff like raw bitpacked, bitpacked in
> > container X etc.
> >
> > Kieran
> > _______________________________________________
> > ffmpeg-devel mailing list
> > ffmpeg-devel at ffmpeg.org
> > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> >
> 
> 
> Didn't know it didn't do that, I guess it's okay then. When its fixed it
> can be unmarked. We still have some experimental features in the psd
> decoder too.

Indeed I added the flag following Kieran's suggestion. Is there anything
else I could do to improve this patchset?

Thanks,
-- 
Damien