[x265] [PATCH] scaling: allow refinement of inter/intra frames when scalefactor is zero

[Kavitha Sampath]

On Mon, Apr 30, 2018 at 11:30 AM, Pradeep Ramachandran <
pradeep at multicorewareinc.com> wrote:

>
>
> On Mon, Apr 30, 2018 at 10:39 AM, Kavitha Sampath <
> kavitha at multicorewareinc.com> wrote:
>
>>
>>
>> On Thu, Apr 26, 2018 at 2:38 PM, Bhavna Hariharan <
>> bhavna at multicorewareinc.com> wrote:
>>
>>>
>>>
>>> On Thu, Apr 26, 2018 at 12:41 PM, <kavitha at multicorewareinc.com> wrote:
>>>
>>>> # HG changeset patch
>>>> # User Kavitha Sampath <kavitha at multicorewareinc.com>
>>>> # Date 1524568273 -19800
>>>> #      Tue Apr 24 16:41:13 2018 +0530
>>>> # Node ID 860c3b32f59e1883b77064b9948e8e17decd4641
>>>> # Parent  07defe235cde1949c55464952ee1dbdc10baf5e4
>>>> scaling: allow refinement of inter/intra frames when scalefactor is zero
>>>>
>>>> Also disable analysis of CU with min-cu size when scale factor is zero
>>>>
>>>> diff -r 07defe235cde -r 860c3b32f59e source/encoder/analysis.cpp
>>>> --- a/source/encoder/analysis.cpp       Thu Apr 12 16:57:19 2018 +0530
>>>> +++ b/source/encoder/analysis.cpp       Tue Apr 24 16:41:13 2018 +0530
>>>> @@ -523,7 +523,7 @@
>>>>      int split = 0;
>>>>      if (m_param->intraRefine && m_param->intraRefine != 4)
>>>>      {
>>>> -        split = ((cuGeom.log2CUSize == (uint32_t)(g_log2Size[m_param->minCUSize]
>>>> + 1)) && bDecidedDepth);
>>>> +        split = m_param->scaleFactor && ((cuGeom.log2CUSize ==
>>>> (uint32_t)(g_log2Size[m_param->minCUSize] + 1)) && bDecidedDepth);
>>>>          if (cuGeom.log2CUSize == (uint32_t)(g_log2Size[m_param->minCUSize])
>>>> && !bDecidedDepth)
>>>>              bAlreadyDecided = false;
>>>>      }
>>>> @@ -2420,7 +2420,7 @@
>>>>          m_refineLevel = m_param->interRefine;
>>>>      else
>>>>          m_refineLevel = m_frame->m_classifyFrame ? 1 : 3;
>>>> -    int split = (m_refineLevel && cuGeom.log2CUSize ==
>>>> (uint32_t)(g_log2Size[m_param->minCUSize] + 1) && bDecidedDepth);
>>>> +    int split = (m_param->scaleFactor && m_refineLevel &&
>>>> cuGeom.log2CUSize == (uint32_t)(g_log2Size[m_param->minCUSize] + 1) &&
>>>> bDecidedDepth);
>>>>      td.split = split;
>>>>
>>>>      if (bDecidedDepth)
>>>> @@ -2494,7 +2494,7 @@
>>>>                              mode.cu.m_mvd[list][pu.puAbsPartIdx] =
>>>> mode.cu.m_mv[list][pu.puAbsPartIdx] - mode.amvpCand[list][ref][mode.
>>>> cu.m_mvpIdx[list][pu.puAbsPartIdx]]/*mvp*/;
>>>>                          }
>>>>                      }
>>>> -                    else if(m_param->scaleFactor)
>>>> +                    else
>>>>                      {
>>>>                          MVField candMvField[MRG_MAX_NUM_CANDS][2]; //
>>>> double length for mv of both lists
>>>>                          uint8_t candDir[MRG_MAX_NUM_CANDS];
>>>> diff -r 07defe235cde -r 860c3b32f59e source/encoder/encoder.cpp
>>>> --- a/source/encoder/encoder.cpp        Thu Apr 12 16:57:19 2018 +0530
>>>> +++ b/source/encoder/encoder.cpp        Tue Apr 24 16:41:13 2018 +0530
>>>> @@ -2694,7 +2694,7 @@
>>>>
>>>>      if (p->intraRefine)
>>>>      {
>>>> -        if (!p->analysisLoad || p->analysisReuseLevel < 10 ||
>>>> !p->scaleFactor)
>>>> +        if (!p->analysisLoad || p->analysisReuseLevel < 10)
>>>>          {
>>>>              x265_log(p, X265_LOG_WARNING, "Intra refinement requires
>>>> analysis load, analysis-reuse-level 10, scale factor. Disabling intra
>>>> refine.\n");
>>>>
>>>
>>> All the warning messages still say that scale factor is required to
>>> enable refinement methods, we can remove that.
>>>
>>
> Can you please fix these warning messages and send an updated patch?
>

Sure. I will send an updated patch

>
>
>>
>>> I see that you've enabled inter-refine, intra-refine and refine-mv
>>> without scale-factor, is there any reason for not enabling dynamic
>>> refinement without scale factor?
>>>
>>
>> Dynamic refinemement + scalefactor 0 requires some more testing. I
>> thought of sending that as a separate patch.
>>
>
> Sounds good.
>
>
>>
>>
>>>
>>>
>>>
>>>>              p->intraRefine = 0;
>>>> @@ -2703,7 +2703,7 @@
>>>>
>>>>      if (p->interRefine)
>>>>      {
>>>> -        if (!p->analysisLoad || p->analysisReuseLevel < 10 ||
>>>> !p->scaleFactor)
>>>> +        if (!p->analysisLoad || p->analysisReuseLevel < 10)
>>>>          {
>>>>              x265_log(p, X265_LOG_WARNING, "Inter refinement requires
>>>> analysis load, analysis-reuse-level 10, scale factor. Disabling inter
>>>> refine.\n");
>>>>              p->interRefine = 0;
>>>> @@ -2737,7 +2737,7 @@
>>>>
>>>>      if (p->mvRefine)
>>>>      {
>>>> -        if (!p->analysisLoad || p->analysisReuseLevel < 10 ||
>>>> !p->scaleFactor)
>>>> +        if (!p->analysisLoad || p->analysisReuseLevel < 10)
>>>>          {
>>>>              x265_log(p, X265_LOG_WARNING, "MV refinement requires
>>>> analysis load, analysis-reuse-level 10, scale factor. Disabling MV
>>>> refine.\n");
>>>>              p->mvRefine = 0;
>>>>
>>>> _______________________________________________
>>>> x265-devel mailing list
>>>> x265-devel at videolan.org
>>>> https://mailman.videolan.org/listinfo/x265-devel
>>>>
>>>>
>>>
>>
>>
>> --
>> Regards,
>> Kavitha
>>
>> _______________________________________________
>> x265-devel mailing list
>> x265-devel at videolan.org
>> https://mailman.videolan.org/listinfo/x265-devel
>>
>>
>
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel
>
>


-- 
Regards,
Kavitha
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20180430/609cd411/attachment.html>