[x265] [PATCH For REVIEW] Bidir ME: new logic adapted from x264

[Steve Borho]

On Mon, Sep 30, 2013 at 8:14 AM, Derek Buitenhuis <
derek.buitenhuis at gmail.com> wrote:

> On 9/30/2013 8:05 AM, deepthidevaki at multicorewareinc.com wrote:
> > # HG changeset patch
> > # User Deepthi Devaki <deepthidevaki at multicorewareinc.com>
> > # Date 1380523193 -19800
> > # Node ID bb238e8b36007aad896884009c720d26df8775c5
> > # Parent  55edc34e253c14d3eccb83a7d1db43774349ff9a
> > Bidir ME: new logic adapted from x264
> >
> > L0 and L1 MVs from unidir ME used for bidir MV. bidir cost is calculated
> from the average of references. Performance/PSNR with new bidir with
> commandline
> > x265.exe FourPeople_1280x720_60.y4m  -f 100 --b-adapt 0 -b 3 --ref 1
> --hash 1 -o four.hevc -r recon.yuv
> > New:  (2.24 fps), 515.16 kb/s, Global PSNR: 39.704
> > Orig: (2.05 fps), 519.47 kb/s, Global PSNR: 39.711
>
> [...]
>
> > -void TEncSearch::predInterSearch(TComDataCU* cu, TComYuv* fencYuv,
> TComYuv* predYuv, bool bUseMRG)
> > +void TEncSearch::predInterSearch(TComDataCU* cu, TComYuv* /*fencYuv*/,
> TComYuv* predYuv, bool bUseMRG)
>
> Um, no.
>
> > +template<int lx, int ly>
> > +void pixelavg_pp(pixel* dst, intptr_t dstride, pixel* src0, pixel*
> src1, intptr_t sstride0, intptr_t sstride1)
> > +{
> > +    for( int y = 0; y < ly; y++ )
> > +    {
> > +        for( int x = 0; x < lx; x++ )
> > +        {
> > +            dst[x] = ( src0[x] + src1[x] + 1 ) >> 1;
> > +        }
> > +        src0 += sstride0;
> > +        src1 += sstride1;
> > +        dst += dstride;
> > +    }
> > +}
>
> Why the template?


It is instantiated for each inter block size (64 in all)

-- 
Steve Borho
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20130930/e2b36e7a/attachment-0001.html>