[FFmpeg-devel] [PATCH] MOV YUV2 fourcc -> PIX_FMT_YUYV422 mapping

[Michael Niedermayer]

On Mon, Mar 30, 2009 at 11:12:10AM +0530, Jai Menon wrote:
> On 3/26/09, Reimar D?ffinger <Reimar.Doeffinger at gmx.de> wrote:
> > On Thu, Mar 26, 2009 at 02:22:03PM +0530, Jai Menon wrote:
> >  > On 3/26/09, Michael Niedermayer <michaelni at gmx.at> wrote:
> >  > > On Wed, Mar 25, 2009 at 08:47:33AM -0700, Baptiste Coudurier wrote:
> >  > >  > On 3/25/2009 8:35 AM, Reimar D?ffinger wrote:
> >  > >  > > On Wed, Mar 25, 2009 at 03:01:50PM +0100, Michael Niedermayer wrote:
> >  > >  > >> On Wed, Mar 25, 2009 at 09:48:51AM +0530, Jai Menon wrote:
> >  > >  > >>> Index: libavcodec/rawdec.c
> >  > >  > >>> ===================================================================
> >  > >  > >>> --- libavcodec/rawdec.c   (revision 17944)
> >  > >  > >>> +++ libavcodec/rawdec.c   (working copy)
> >  > >  > >>> @@ -144,6 +144,13 @@
> >  > >  > >>>          picture->data[2] = tmp;
> >  > >  > >>>      }
> >  > >  > >>>
> >  > >  > >>> +    if(avctx->codec_tag == MKTAG('y', 'u', 'v', '2')) {
> >  > >  > >>> +        int i;
> >  > >  > >>> +        for(i=1; i<picture->linesize[0]*avctx->height; i+=2) {
> >  > >  > >>> +            picture->data[0][i] ^= 128;
> >  > >  > >>> +        }
> >  > >  > >>> +    }
> >
> >
> > I think you should avoid writing to the padding pixels, particularly
> >  since this would not work for negative linesize i.e. something
> >  like
> >  uint8_t *line = picture->data[0];
> >  for (y = 0; y < avctx->height; y++) {
> >     for (x = 0; x < avctx->width; x++)
> >         line[2*x + 1] ^= 0x80; // convert signed to unsigned
> >     line += picture->linesize[0];
> >
> > }
> >
> 
> Changed. I didn't realize linesize could be negative.
> 
> >  > Are you suggesting adding a similar if case in raw_encode which sets
> >  > the sign bit? such a patch will be quite intrusive, especially for a
> >  > pixel fmt which most people won't use :)
> >
> >
> > Huh? Wouldn't you just need to add exactly the same code between
> >  avpicture_layout and the return?
> 
> Nevermind, I was looking at adding it in imgconvert code :)
> Revised patch attached.
> 
> -- 
> Regards,
> 
> Jai

> Index: libavcodec/rawenc.c
> ===================================================================
> --- libavcodec/rawenc.c	(revision 17944)
> +++ libavcodec/rawenc.c	(working copy)
> @@ -40,8 +40,19 @@
>  static int raw_encode(AVCodecContext *avctx,
>                              unsigned char *frame, int buf_size, void *data)
>  {
> -    return avpicture_layout((AVPicture *)data, avctx->pix_fmt, avctx->width,
> +    int ret = avpicture_layout((AVPicture *)data, avctx->pix_fmt, avctx->width,
>                                                 avctx->height, frame, buf_size);
> +
> +    if(avctx->codec_tag == MKTAG('y', 'u', 'v', '2') && (ret > 0)) {

doesnt this need a pix_fmt check too?


> +        int i, j;
> +        uint8_t *line = frame;
> +        for(j = 0; j < avctx->height; j++) {
> +            for(i = 0; i < avctx->width; i++)

x and y are better variable names in this case

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

If you really think that XML is the answer, then you definitly missunderstood
the question -- Attila Kinali
-------------- 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/20090330/4f5e46e4/attachment.pgp>