[FFmpeg-devel] MXF D10 regression tests

Reimar Döffinger Reimar.Doeffinger
Wed Mar 18 14:46:48 CET 2009


On Wed, Mar 18, 2009 at 02:31:33PM +0100, Michael Niedermayer wrote:
> > @@ -243,7 +243,7 @@
> >          int size[4] = {0};
> >          int tmpsize;
> >          AVPicture picture;
> > -        int stride_align[4];
> > +        int stride_align = STRIDE_ALIGN;
> >  
> >          avcodec_get_chroma_sub_sample(s->pix_fmt, &h_chroma_shift, &v_chroma_shift);
> >  
> 
> this is conceptually slightly bad because
> 16x16 MBs in luma may need 16 byte alignment and
> 8x8 MBs in 420 chroma may need 8 byte alignment

Current code does not care, it just saves a rather small amount of
memory and it adds a bit of complexity.
Not to mention that there might be quite a few codecs that need no
alignment at all. I am rather unconvinced that there is a point in
taking all this into account.




More information about the ffmpeg-devel mailing list