[x265] [PATCH] cutree: identify the correct b-ref to estimate propagatecost

[Steve Borho]

On Thu, May 15, 2014 at 2:02 AM,  <gopu at multicorewareinc.com> wrote:
> # HG changeset patch
> # User Gopu Govindaswamy
> # Date 1400137266 -19800
> #      Thu May 15 12:31:06 2014 +0530
> # Node ID 6ca880d7e68361e5b9e2353f05369d9e5c40a29f
> # Parent  d5b42a9fe43ba877f90f73a11a071d9e449e9661
> cutree: identify the correct b-ref to estimate propagatecost

This patch is blocked until I get a response from the x264 developers
about the same issue in their slicetype.c

>
> diff -r d5b42a9fe43b -r 6ca880d7e683 source/encoder/slicetype.cpp
> --- a/source/encoder/slicetype.cpp      Wed May 14 21:01:56 2014 +0530
> +++ b/source/encoder/slicetype.cpp      Thu May 15 12:31:06 2014 +0530
> @@ -974,7 +974,7 @@
>          bframes = lastnonb - curnonb - 1;
>          if (param->bBPyramid && bframes > 1)
>          {
> -            int middle = (bframes + 1) / 2 + curnonb;
> +            int middle = bframes / 2 + curnonb;
>              est.estimateFrameCost(frames, curnonb, lastnonb, middle, 0);
>              memset(frames[middle]->propagateCost, 0, cuCount * sizeof(uint16_t));
>              while (i > curnonb)
> @@ -1013,7 +1013,7 @@
>
>      cuTreeFinish(frames[lastnonb], averageDuration, lastnonb);
>      if (param->bBPyramid && bframes > 1 && !param->rc.vbvBufferSize)
> -        cuTreeFinish(frames[lastnonb + (bframes + 1) / 2], averageDuration, 0);
> +        cuTreeFinish(frames[lastnonb + bframes / 2], averageDuration, 0);
>  }
>
>  void Lookahead::estimateCUPropagate(Lowres **frames, double averageDuration, int p0, int p1, int b, int referenced)
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel



-- 
Steve Borho