[x265] [PATCH] bug fix: allow split in frame boundary for scaled encodes

[Ashok Kumar Mishra]

On Tue, May 29, 2018 at 7:23 PM, <bhavna at multicorewareinc.com> wrote:

> # HG changeset patch
> # User Bhavna Hariharan <bhavna at multicorewareinc.com>
> # Date 1527578404 -19800
> #      Tue May 29 12:50:04 2018 +0530
> # Node ID 9db5be89d74cc55cd6aff30c2a576119ca09d8a1
> # Parent  9389296cd844bf14850b98ee181772be94d209b9
> bug fix: allow split in frame boundary for scaled encodes.
>
> For cases where the entire CTU is not within the frame boundary, the CU
> must be
> allowed to split when the depth from the save encode is not applicable.
>
> diff -r 9389296cd844 -r 9db5be89d74c source/encoder/analysis.cpp
> --- a/source/encoder/analysis.cpp       Mon May 28 14:55:57 2018 +0530
> +++ b/source/encoder/analysis.cpp       Tue May 29 12:50:04 2018 +0530
> @@ -523,14 +523,15 @@
>      int split = 0;
>      if (m_param->intraRefine && m_param->intraRefine != 4)
>      {
> -        split = m_param->scaleFactor && ((cuGeom.log2CUSize ==
> (uint32_t)(g_log2Size[m_param->minCUSize] + 1)) && bDecidedDepth);
> +        split = m_param->scaleFactor && bDecidedDepth && (!mightNotSplit
> ||
> +            ((cuGeom.log2CUSize == (uint32_t)(g_log2Size[m_param->minCUSize]
> + 1))));
>          if (cuGeom.log2CUSize == (uint32_t)(g_log2Size[m_param->minCUSize])
> && !bDecidedDepth)
>              bAlreadyDecided = false;
>      }
>
>      if (bAlreadyDecided)
>      {
> -        if (bDecidedDepth)
> +        if (bDecidedDepth && mightNotSplit)
>          {
>              Mode& mode = md.pred[0];
>              md.bestMode = &mode;
> @@ -2420,10 +2421,11 @@
>          m_refineLevel = m_param->interRefine;
>      else
>          m_refineLevel = m_frame->m_classifyFrame ? 1 : 3;
> -    int split = (m_param->scaleFactor && m_refineLevel &&
> cuGeom.log2CUSize == (uint32_t)(g_log2Size[m_param->minCUSize] + 1) &&
> bDecidedDepth);
> +    int split = (m_param->scaleFactor && bDecidedDepth && (!mightNotSplit
> ||
> +        (m_refineLevel && cuGeom.log2CUSize ==
> (uint32_t)(g_log2Size[m_param->minCUSize] + 1))));
>      td.split = split;
>
> -    if (bDecidedDepth)
> +    if (bDecidedDepth && mightNotSplit)
>      {
>          setLambdaFromQP(parentCTU, qp, lqp);
>
>
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel
>
>
Pushed.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20180529/c2314fb0/attachment.html>