[x265] [PATCH] Disable QP limitations for lowpass subband dct

Pradeep Ramachandran pradeep at multicorewareinc.com
Wed Nov 8 06:46:49 CET 2017


On Wed, Nov 8, 2017 at 8:44 AM, <mont3z.claro5 at gmail.com> wrote:

> # HG changeset patch
> # User hribeiro
> # Date 1510110049 28800
> #      Tue Nov 07 19:00:49 2017 -0800
> # Node ID e7e951834b57633ae474f50e7878b24271afabd2
> # Parent  b6c692f5f5b145eb8d874cfb44db8c85fc2978cc
> Disable QP limitations for lowpass subband dct
>

Thanks. Temporarily, I agree that removing the requirement to do support
lowpass-dct only with CQP>23 is good.
I've slightly adjusted the docs below and pushed this to default branch.

We need to figure out if we can do this dynamically based on the content.


>
> diff -r b6c692f5f5b1 -r e7e951834b57 doc/reST/cli.rst
> --- a/doc/reST/cli.rst  Tue Nov 07 19:15:31 2017 +0530
> +++ b/doc/reST/cli.rst  Tue Nov 07 19:00:49 2017 -0800
> @@ -2165,10 +2165,10 @@
>
>  .. option:: --lowpass-dct
>
> -    If enabled, x265 will use low-pass truncated dct approximation
> instead of the
> -    standard dct. This approximation is less computational intesive but
> it generates
> -    truncated coefficient matrixes for the transformed block. Empirical
> analysis shows
> -    this approximation gives good PSNR results for QP>=23.
> +    If enabled, x265 will use low-pass subband dct approximation instead
> of the
> +    standard dct for 16x16 and 32x32 blocks. This approximation is less
> computational
> +    intesive but it generates truncated coefficient matrixes for the
> transformed block.
> +    Empirical analysis shows marginal loss in compression and performance
> gains up to 10%.
>
>      This approximation should be considered for platforms with
> performance and time
>      constrains.
> diff -r b6c692f5f5b1 -r e7e951834b57 source/common/primitives.cpp
> --- a/source/common/primitives.cpp      Tue Nov 07 19:15:31 2017 +0530
> +++ b/source/common/primitives.cpp      Tue Nov 07 19:00:49 2017 -0800
> @@ -272,7 +272,7 @@
>
>          setupAliasPrimitives(primitives);
>
> -        if (param->bLowPassDct && param->rc.qp > 20)
> +        if (param->bLowPassDct)
>          {
>              enableLowpassDCTPrimitives(primitives);
>          }
> diff -r b6c692f5f5b1 -r e7e951834b57 source/x265.h
> --- a/source/x265.h     Tue Nov 07 19:15:31 2017 +0530
> +++ b/source/x265.h     Tue Nov 07 19:00:49 2017 -0800
> @@ -1510,9 +1510,8 @@
>      /* Disable lookahead */
>      int       bDisableLookahead;
>
> -    /* Use low-pass truncated dct approximation
> -    *  This DCT approximation is less computational intensive and gives
> results close to
> -    *  standard DCT for QP >= 23 */
> +    /* Use low-pass subband dct approximation
> +    *  This DCT approximation is less computational intensive and gives
> results close to standard DCT */
>      int       bLowPassDct;
>
>      /* Sets the portion of the decode buffer that must be available after
> all the
>
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20171108/62d5978e/attachment.html>


More information about the x265-devel mailing list