[Ffmpeg-devel] [PATCH/RFC] 1-7 and 9-15 bits per pixel PGM files

[Michael Niedermayer]

Hi

On Sun, Apr 08, 2007 at 07:42:54PM +0200, Ivo wrote:
> Hi,
> 
> On Sunday 08 April 2007 18:27, Michael Niedermayer wrote:
> > On Sun, Apr 08, 2007 at 02:30:36PM +0200, Ivo wrote:
> > > On Sunday 08 April 2007 14:25, Ivo wrote:
> > > > Attached patch incorporates your suggestion.
> > >
> > > And without the superfluous {}'s
> >
> > why are the 16 and 8 bit cases done differently?
> > that is they use differnt rounding behavior ...
> > IMHO it should be consistent
> 
> Yes, I agree. Here's another patch.
> 
> --Ivo

> Index: libavcodec/pnm.c
> ===================================================================
> --- libavcodec/pnm.c	(revision 8648)
> +++ libavcodec/pnm.c	(working copy)
> @@ -26,6 +26,7 @@
>      uint8_t *bytestream_start;
>      uint8_t *bytestream_end;
>      AVFrame picture;
> +    uint16_t maxval;             ///< maximum value of a pixel
>  } PNMContext;

why not int?


[...]
> +            for(j = 0; j < avctx->width; j++) {
> +                val = AV_RB16(&ptr[j<<1]);
> +                ((uint16_t *)ptr)[j] = (val * f + 16384)>>15;

this has one fatal problem, mixing reads and writes of different size to the
same area in memory is slow on some cpus (AV_RB16() can use bytewise reads)
reading uint16_t style and using be2me_16() would avoid that

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

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070408/8c9c0fb0/attachment.pgp>