[FFmpeg-devel] [PATCH] PB-frames support for i263

[Michael Niedermayer]

On Wed, Feb 18, 2009 at 10:40:56AM +0200, Kostya wrote:
> On Tue, Feb 17, 2009 at 06:48:57PM +0100, Michael Niedermayer wrote:
> > On Tue, Feb 17, 2009 at 08:29:36PM +0200, Kostya wrote:
> [..]
> > > separated
> > > Since i263 are the only codec known (to me) to contain PB-frames, it's the
> > > only way to test code.
> > 
> > huh? ITU h263+ allows PB frames, and id be surprised if there are no reference
> > streams around ...
> 
> I'd like to find them too.

download tmn and generate your own


> 
> [skipping B-part of PB-frame in ff_h263_decode_mb()]
> > 
> > the code above that alraedy was there also can be factored out
> > also s->mb_intra = 0; can always be done before and the one afterwards
> > is just restoring the overwritten mode i suspect
> 
> It is. The problem is factoring out that code - PB-frame P-part may contain
> both intra- and interblocks, and B-part stuff is dispersed between them
> (i.e. own CBP before P-part CBP, motion vectors after P-part motion vectors,
> own block data after I- or P-block data).
> 
> [PB-frame support in I263]
> > i dont think that will compile without the other patch and the other
> > patch will not work without that
> > aka poorly split
>  
> What about this patch for better parsing of i263 frame header?
>  
> > [...]
> > -- 
> > Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

> Index: libavcodec/h263.c
> ===================================================================
> --- libavcodec/h263.c	(revision 16921)
> +++ libavcodec/h263.c	(working copy)
> @@ -6185,17 +6232,54 @@
>          return -1;      /* SAC: off */
>      }
>      s->obmc= get_bits1(&s->gb);
> -    if (get_bits1(&s->gb) != 0) {
> -        av_log(s->avctx, AV_LOG_ERROR, "PB frame mode no supported\n");
> -        return -1;      /* PB frame mode */
> +    s->pb_frame = get_bits1(&s->gb);
> +
> +    if(format == 7){
> +        format = get_bits(&s->gb, 3);
> +        if(format == 0 || format == 7){
> +            av_log(s->avctx, AV_LOG_ERROR, "Wrong Intel H263 format\n");
> +            return -1;
> +        }
> +        if(get_bits(&s->gb, 2)){
> +            av_log(s->avctx, AV_LOG_ERROR, "Bad value for reserved field\n");

> +            return -1;

dont return here, just print an error


> +        }
> +        s->loop_filter = get_bits1(&s->gb);
> +        if(get_bits1(&s->gb)){
> +            av_log(s->avctx, AV_LOG_ERROR, "Bad value for reserved field\n");
> +            return -1;

dont return here, just print an error


> +        }
> +        if(get_bits1(&s->gb))
> +            s->pb_frame = 2;
> +        if(get_bits(&s->gb, 5)){
> +            av_log(s->avctx, AV_LOG_ERROR, "Bad value for reserved field\n");
> +            return -1;

dont return here, just print an error


> +        }
> +        if(get_bits(&s->gb, 5) != 1){
> +            av_log(s->avctx, AV_LOG_ERROR, "Invalid marker\n");
> +            return -1;

dont return here, just print an error


> +        }
>      }
> +    if(format == 6){
> +        int ar = get_bits(&s->gb, 4);
> +        skip_bits(&s->gb, 9); // display width

> +        if(!get_bits1(&s->gb))
> +            return -1;

dont return here, just print an error


[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

> ... defining _GNU_SOURCE...
For the love of all that is holy, and some that is not, don't do that.
-- Luca & Mans
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090218/77367316/attachment.pgp>