ref: fb4c907ecbb3c4af9ac372060a55efa1f0de1f11
dir: /codec/encoder/core/src/svc_mode_decision.cpp/
/*!
* \copy
* Copyright (c) 2009-2013, Cisco Systems
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
*
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
* COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*
*
* \file svc_mode_decision.c
*
* \brief Algorithmetic MD for:
* - multi-spatial Enhancement Layer MD;
* - Scrolling PSkip Decision for screen content
*
* \date 2009.7.29
*
**************************************************************************************
*/
#include "mv_pred.h"
#include "ls_defines.h"
#include "svc_base_layer_md.h"
#include "svc_mode_decision.h"
namespace WelsSVCEnc {
//////////////
// MD for enhancement layers
//////////////
void WelsMdSpatialelInterMbIlfmdNoilp (sWelsEncCtx* pEncCtx, SWelsMD* pWelsMd, SSlice* pSlice,
SMB* pCurMb, const Mb_Type kuiRefMbType) {
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
SMbCache* pMbCache = &pSlice->sMbCacheInfo;
const uint32_t kuiNeighborAvail = pCurMb->uiNeighborAvail;
const int32_t kiMbWidth = pCurDqLayer->iMbWidth;
const SMB* kpTopMb = pCurMb - kiMbWidth;
const bool kbMbLeftAvailPskip = ((kuiNeighborAvail & LEFT_MB_POS) ? IS_SKIP ((pCurMb - 1)->uiMbType) : false);
const bool kbMbTopAvailPskip = ((kuiNeighborAvail & TOP_MB_POS) ? IS_SKIP (kpTopMb->uiMbType) : false);
const bool kbMbTopLeftAvailPskip = ((kuiNeighborAvail & TOPLEFT_MB_POS) ? IS_SKIP ((kpTopMb - 1)->uiMbType) : false);
const bool kbMbTopRightAvailPskip = ((kuiNeighborAvail & TOPRIGHT_MB_POS) ? IS_SKIP ((
kpTopMb + 1)->uiMbType) : false);
bool bTrySkip = kbMbLeftAvailPskip | kbMbTopAvailPskip | kbMbTopLeftAvailPskip | kbMbTopRightAvailPskip;
bool bKeepSkip = kbMbLeftAvailPskip & kbMbTopAvailPskip & kbMbTopRightAvailPskip;
bool bSkip = false;
if (pEncCtx->pFuncList->pfInterMdBackgroundDecision (pEncCtx, pWelsMd, pSlice, pCurMb, pMbCache, &bKeepSkip)) {
return;
}
//step 1: try SKIP
bSkip = WelsMdInterJudgePskip (pEncCtx, pWelsMd, pSlice, pCurMb, pMbCache, bTrySkip);
if (bSkip && bKeepSkip) {
WelsMdInterDecidedPskip (pEncCtx, pSlice, pCurMb, pMbCache);
return;
}
if (! IS_SVC_INTRA (kuiRefMbType)) {
if (!bSkip) {
PredictSad (pMbCache->sMvComponents.iRefIndexCache, pMbCache->iSadCost, 0, &pWelsMd->iSadPredMb);
//step 2: P_16x16
pWelsMd->iCostLuma = WelsMdP16x16 (pEncCtx->pFuncList, pCurDqLayer, pWelsMd, pSlice, pCurMb);
pCurMb->uiMbType = MB_TYPE_16x16;
}
WelsMdInterSecondaryModesEnc (pEncCtx, pWelsMd, pSlice, pCurMb, pMbCache, bSkip);
} else { //BLMODE == SVC_INTRA
//initial prediction memory for I_16x16
const int32_t kiCostI16x16 = WelsMdI16x16 (pEncCtx->pFuncList, pEncCtx->pCurDqLayer, pMbCache, pWelsMd->iLambda);
if (bSkip && (pWelsMd->iCostLuma <= kiCostI16x16)) {
WelsMdInterDecidedPskip (pEncCtx, pSlice, pCurMb, pMbCache);
} else {
pWelsMd->iCostLuma = kiCostI16x16;
pCurMb->uiMbType = MB_TYPE_INTRA16x16;
WelsMdIntraSecondaryModesEnc (pEncCtx, pWelsMd, pCurMb, pMbCache);
}
}
}
void WelsMdInterMbEnhancelayer (void* pEnc, void* pMd, SSlice* pSlice, SMB* pCurMb, SMbCache* pMbCache) {
sWelsEncCtx* pEncCtx = (sWelsEncCtx*)pEnc;
SDqLayer* pCurLayer = pEncCtx->pCurDqLayer;
SWelsMD* pWelsMd = (SWelsMD*)pMd;
const SMB* kpInterLayerRefMb = GetRefMb (pCurLayer, pCurMb);
const Mb_Type kuiInterLayerRefMbType = kpInterLayerRefMb->uiMbType;
SetMvBaseEnhancelayer (pWelsMd, pCurMb,
kpInterLayerRefMb); // initial sMvBase here only when pRef mb type is inter, if not sMvBase will be not used!
//step (3): do the MD process
WelsMdSpatialelInterMbIlfmdNoilp (pEncCtx, pWelsMd, pSlice, pCurMb, kuiInterLayerRefMbType); //MD process
}
// do initiation for noILP (needed by ILFMD)
SMB* GetRefMb (SDqLayer* pCurLayer, SMB* pCurMb) {
const SDqLayer* kpRefLayer = pCurLayer->pRefLayer;
const int32_t kiRefMbIdx = (pCurMb->iMbY >> 1) * kpRefLayer->iMbWidth + (pCurMb->iMbX >>
1); //because current lower layer is half size on both vertical and horizontal
return (&kpRefLayer->sMbDataP[kiRefMbIdx]);
}
void SetMvBaseEnhancelayer (SWelsMD* pMd, SMB* pCurMb, const SMB* kpRefMb) {
const Mb_Type kuiRefMbType = kpRefMb->uiMbType;
if (! IS_SVC_INTRA (kuiRefMbType)) {
SMVUnitXY sMv;
int32_t iRefMbPartIdx = ((pCurMb->iMbY & 0x01) << 1) + (pCurMb->iMbX & 0x01); //may be need modified
int32_t iScan4RefPartIdx = g_kuiMbCountScan4Idx[ (iRefMbPartIdx << 2)];
sMv.iMvX = kpRefMb->sMv[iScan4RefPartIdx].iMvX << 1;
sMv.iMvY = kpRefMb->sMv[iScan4RefPartIdx].iMvY << 1;
pMd->sMe.sMe16x16.sMvBase = sMv;
pMd->sMe.sMe8x8[0].sMvBase =
pMd->sMe.sMe8x8[1].sMvBase =
pMd->sMe.sMe8x8[2].sMvBase =
pMd->sMe.sMe8x8[3].sMvBase = sMv;
pMd->sMe.sMe16x8[0].sMvBase =
pMd->sMe.sMe16x8[1].sMvBase =
pMd->sMe.sMe8x16[0].sMvBase =
pMd->sMe.sMe8x16[1].sMvBase = sMv;
}
}
//////////////
// MD for Background decision
//////////////
//////
// try the BGD Pskip
//////
inline int32_t GetChromaCost (PSampleSadSatdCostFunc* pCalculateFunc,
uint8_t* pSrcChroma, int32_t iSrcStride, uint8_t* pRefChroma, int32_t iRefStride) {
return pCalculateFunc[BLOCK_8x8] (pSrcChroma, iSrcStride, pRefChroma, iRefStride);
}
inline bool IsCostLessEqualSkipCost (int32_t iCurCost, const int32_t iPredPskipSad, const int32_t iRefMbType,
const SPicture* pRef, const int32_t iMbXy, const int32_t iSmallestInvisibleTh) {
return ((iPredPskipSad > iSmallestInvisibleTh && iCurCost >= iPredPskipSad) ||
(pRef->iPictureType == P_SLICE &&
iRefMbType == MB_TYPE_SKIP &&
pRef->pMbSkipSad[iMbXy] > iSmallestInvisibleTh &&
iCurCost >= (pRef->pMbSkipSad[iMbXy])));
}
bool CheckChromaCost (sWelsEncCtx* pEncCtx, SWelsMD* pWelsMd, SMbCache* pMbCache, const int32_t iCurMbXy) {
#define KNOWN_CHROMA_TOO_LARGE 640
#define SMALLEST_INVISIBLE 128 //2*64, 2 in pixel maybe the smallest not visible for luma
PSampleSadSatdCostFunc* pSad = pEncCtx->pFuncList->sSampleDealingFuncs.pfSampleSad;
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
uint8_t* pCbEnc = pMbCache->SPicData.pEncMb[1];
uint8_t* pCrEnc = pMbCache->SPicData.pEncMb[2];
uint8_t* pCbRef = pMbCache->SPicData.pRefMb[1];
uint8_t* pCrRef = pMbCache->SPicData.pRefMb[2];
const int32_t iCbEncStride = pCurDqLayer->iEncStride[1];
const int32_t iCrEncStride = pCurDqLayer->iEncStride[2];
const int32_t iChromaRefStride = pCurDqLayer->pRefPic->iLineSize[1];
const int32_t iCbSad = GetChromaCost (pSad, pCbEnc, iCbEncStride, pCbRef, iChromaRefStride);
const int32_t iCrSad = GetChromaCost (pSad, pCrEnc, iCrEncStride, pCrRef, iChromaRefStride);
//01/17/13
//the in-question error area is
//from: (yellow) Y=212, V=023, U=145
//to: (grey) Y=213, V=136, U=124
//visible difference can be seen on the U plane
//so the allowing chroma difference should be at least no larger than
//20*8*8 = 1280 for U or V
//one local test case show that "either one >640" will become a too strict criteria, which will appear when QP is large(36) and maybe no much harm for visual
//another local test case show that "either one >960" will be a moderate criteria, an area is changed from light green to light pink, but without careful observation it won't be obvious, but people will feel the unclean area (and note that, the color visible criteria is also related to the luma of them!)
//another case show that color changed from black to very dark red can be visible even under the threshold 960, the color difference is about 13*64=832 (U123V145->U129V132)
//TODO:
//OPTI-ABLE: the visible color criteria may be related to luma (very bright or very dark), or related to the ratio of U/V rather than the absolute value
const bool bChromaTooLarge = (iCbSad > KNOWN_CHROMA_TOO_LARGE || iCrSad > KNOWN_CHROMA_TOO_LARGE);
const int32_t iChromaSad = iCbSad + iCrSad;
PredictSadSkip (pMbCache->sMvComponents.iRefIndexCache, pMbCache->bMbTypeSkip, pMbCache->iSadCostSkip, 0,
& (pWelsMd->iSadPredSkip));
const bool bChromaCostCannotSkip = IsCostLessEqualSkipCost (iChromaSad, pWelsMd->iSadPredSkip, pMbCache->uiRefMbType,
pCurDqLayer->pRefPic, iCurMbXy, SMALLEST_INVISIBLE);
return (!bChromaCostCannotSkip && !bChromaTooLarge);
}
//01/17/2013. USE the NEW BGD Pskip with COLOR CHECK for screen content and camera because of color artifact seen in test
bool WelsMdInterJudgeBGDPskip (void* pCtx, void* pMd, SSlice* pSlice, SMB* pCurMb, SMbCache* pMbCache,
bool* bKeepSkip) {
sWelsEncCtx* pEncCtx = (sWelsEncCtx*)pCtx;
SWelsMD* pWelsMd = (SWelsMD*)pMd;
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
const int32_t kiRefMbQp = pCurDqLayer->pRefPic->pRefMbQp[pCurMb->iMbXY];
const int32_t kiCurMbQp = pCurMb->uiLumaQp;// unsigned -> signed
int8_t* pVaaBgMbFlag = pEncCtx->pVaa->pVaaBackgroundMbFlag + pCurMb->iMbXY;
const int32_t kiMbWidth = pCurDqLayer->iMbWidth;
*bKeepSkip = (*bKeepSkip) &&
((!pVaaBgMbFlag[-1]) &&
(!pVaaBgMbFlag[-kiMbWidth]) &&
(!pVaaBgMbFlag[-kiMbWidth + 1]));
if (
*pVaaBgMbFlag
&& !IS_INTRA (pMbCache->uiRefMbType)
&& (kiRefMbQp - kiCurMbQp <= DELTA_QP_BGD_THD || kiRefMbQp <= 26)
) {
//01/16/13
//the current BGD method uses luma SAD in first step judging of Background blocks
//and uses chroma edges to confirm the Background blocks
//HOWEVER, there is such case in SCC,
//that the luma of two collocated blocks (block in reference frame and in current frame) is very similar
//but the chroma are very different, at the same time the chroma are plain and without edge
//IN SUCH A CASE,
//it will be not proper to just use Pskip
//TODO: consider reusing this result of ChromaCheck when SCDSkip needs this as well
if (CheckChromaCost (pEncCtx, pWelsMd, pMbCache, pCurMb->iMbXY)) {
SMVUnitXY sVaaPredSkipMv = { 0 };
PredSkipMv (pMbCache, &sVaaPredSkipMv);
WelsMdBackgroundMbEnc (pEncCtx, pWelsMd, pCurMb, pMbCache, pSlice, (LD32 (&sVaaPredSkipMv) == 0));
return true;
}
}
return false;
}
bool WelsMdInterJudgeBGDPskipFalse (void* pCtx, void* pMd, SSlice* pSlice, SMB* pCurMb, SMbCache* pMbCache,
bool* bKeepSkip) {
return false;
}
//////
// update BGD related info
//////
void WelsMdInterUpdateBGDInfo (SDqLayer* pCurLayer, SMB* pCurMb, const bool bCollocatedPredFlag,
const int32_t iRefPictureType) {
uint8_t* pTargetRefMbQpList = (pCurLayer->pDecPic->pRefMbQp);
const int32_t kiMbXY = pCurMb->iMbXY;
if (pCurMb->uiCbp || I_SLICE == iRefPictureType || 0 == bCollocatedPredFlag) {
pTargetRefMbQpList[kiMbXY] = pCurMb->uiLumaQp;
} else { //unchange, do not need to evaluation?
uint8_t* pRefPicRefMbQpList = (pCurLayer->pRefPic->pRefMbQp);
pTargetRefMbQpList[kiMbXY] = pRefPicRefMbQpList[kiMbXY];
}
if (pCurMb->uiMbType == MB_TYPE_BACKGROUND) {
pCurMb->uiMbType = MB_TYPE_SKIP;
}
}
void WelsMdInterUpdateBGDInfoNULL (SDqLayer* pCurLayer, SMB* pCurMb, const bool bCollocatedPredFlag,
const int32_t iRefPictureType) {
}
//////////////
// MD for screen contents
//////////////
inline bool IsMbStatic (int32_t* pBlockType, EStaticBlockIdc eType) {
return (pBlockType != NULL &&
eType == pBlockType[0] &&
eType == pBlockType[1] &&
eType == pBlockType[2] &&
eType == pBlockType[3]);
}
inline bool IsMbCollocatedStatic (int32_t* pBlockType) {
return IsMbStatic (pBlockType, COLLOCATED_STATIC);
}
inline bool IsMbScrolledStatic (int32_t* pBlockType) {
return IsMbStatic (pBlockType, SCROLLED_STATIC);
}
inline int32_t CalUVSadCost (SWelsFuncPtrList* pFunc, uint8_t* pEncOri, int32_t iStrideUV, uint8_t* pRefOri,
int32_t iRefLineSize) {
return pFunc->sSampleDealingFuncs.pfSampleSad[BLOCK_8x8] (pEncOri, iStrideUV, pRefOri, iRefLineSize);
}
inline bool CheckBorder (int32_t iMbX, int32_t iMbY, int32_t iScrollMvX, int32_t iScrollMvY, int32_t iMbWidth,
int32_t iMbHeight) {
return ((iMbX << 4) + iScrollMvX < 0 ||
(iMbX << 4) + iScrollMvX > (iMbWidth - 1) << 4 ||
(iMbY << 4) + iScrollMvY < 0 ||
(iMbY << 4) + iScrollMvY > (iMbHeight - 1) << 4
); //border check for safety
}
bool JudgeStaticSkip (sWelsEncCtx* pEncCtx, SMB* pCurMb, SMbCache* pMbCache, SWelsMD* pWelsMd) {
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
const int32_t kiMbX = pCurMb->iMbX;
const int32_t kiMbY = pCurMb->iMbY;
bool bTryStaticSkip = IsMbCollocatedStatic (pWelsMd->iBlock8x8StaticIdc);
if (bTryStaticSkip) {
int32_t iStrideUV, iOffsetUV;
SWelsFuncPtrList* pFunc = pEncCtx->pFuncList;
SPicture* pRefOri = pCurDqLayer->pRefOri[0];
if (pRefOri != NULL) {
iStrideUV = pCurDqLayer->iEncStride[1];
iOffsetUV = (kiMbX + kiMbY * iStrideUV) << 3;
int32_t iSadCostCb = CalUVSadCost (pFunc, pMbCache->SPicData.pEncMb[1], iStrideUV, pRefOri->pData[1] + iOffsetUV,
pRefOri->iLineSize[1]);
if (iSadCostCb == 0) {
int32_t iSadCostCr = CalUVSadCost (pFunc, pMbCache->SPicData.pEncMb[2], iStrideUV, pRefOri->pData[2] + iOffsetUV,
pRefOri->iLineSize[1]);
bTryStaticSkip = (0 == iSadCostCr);
} else bTryStaticSkip = false;
}
}
return bTryStaticSkip;
}
bool JudgeScrollSkip (sWelsEncCtx* pEncCtx, SMB* pCurMb, SMbCache* pMbCache, SWelsMD* pWelsMd) {
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
const int32_t kiMbX = pCurMb->iMbX;
const int32_t kiMbY = pCurMb->iMbY;
const int32_t kiMbWidth = pCurDqLayer->iMbWidth;
const int32_t kiMbHeight = pCurDqLayer->iMbHeight;
// const int32_t block_width = mb_width << 1;
SVAAFrameInfoExt_t* pVaaExt = static_cast<SVAAFrameInfoExt_t*> (pEncCtx->pVaa);
bool bTryScrollSkip = false;
if (pVaaExt->sScrollDetectInfo.bScrollDetectFlag)
bTryScrollSkip = IsMbScrolledStatic (pWelsMd->iBlock8x8StaticIdc);
else return 0;
if (bTryScrollSkip) {
int32_t iStrideUV, iOffsetUV;
SWelsFuncPtrList* pFunc = pEncCtx->pFuncList;
SPicture* pRefOri = pCurDqLayer->pRefOri[0];
if (pRefOri != NULL) {
int32_t iScrollMvX = pVaaExt->sScrollDetectInfo.iScrollMvX;
int32_t iScrollMvY = pVaaExt->sScrollDetectInfo.iScrollMvY;
if (CheckBorder (kiMbX, kiMbY, iScrollMvX, iScrollMvY, kiMbWidth, kiMbHeight)) {
bTryScrollSkip = false;
} else {
iStrideUV = pCurDqLayer->iEncStride[1];
iOffsetUV = (kiMbX << 3) + (iScrollMvX >> 1) + ((kiMbY << 3) + (iScrollMvY >> 1)) * iStrideUV;
int32_t iSadCostCb = CalUVSadCost (pFunc, pMbCache->SPicData.pEncMb[1], iStrideUV, pRefOri->pData[1] + iOffsetUV,
pRefOri->iLineSize[1]);
if (iSadCostCb == 0) {
int32_t iSadCostCr = CalUVSadCost (pFunc, pMbCache->SPicData.pEncMb[2], iStrideUV, pRefOri->pData[2] + iOffsetUV,
pRefOri->iLineSize[1]);
bTryScrollSkip = (0 == iSadCostCr);
} else bTryScrollSkip = false;
}
}
}
return bTryScrollSkip;
}
void SvcMdSCDMbEnc (sWelsEncCtx* pEncCtx, SWelsMD* pWelsMd, SMB* pCurMb, SMbCache* pMbCache, SSlice* pSlice,
bool bQpSimilarFlag,
bool bMbSkipFlag, SMVUnitXY sCurMbMv[], ESkipModes eSkipMode) {
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
SWelsFuncPtrList* pFunc = pEncCtx->pFuncList;
SMVUnitXY sMvp = { 0};
ST16 (&sMvp.iMvX, sCurMbMv[eSkipMode].iMvX);
ST16 (&sMvp.iMvY, sCurMbMv[eSkipMode].iMvY);
uint8_t* pRefLuma = pMbCache->SPicData.pRefMb[0];
uint8_t* pRefCb = pMbCache->SPicData.pRefMb[1];
uint8_t* pRefCr = pMbCache->SPicData.pRefMb[2];
int32_t iLineSizeY = pCurDqLayer->pRefPic->iLineSize[0];
int32_t iLineSizeUV = pCurDqLayer->pRefPic->iLineSize[1];
uint8_t* pDstLuma = pMbCache->pSkipMb;
uint8_t* pDstCb = pMbCache->pSkipMb + 256;
uint8_t* pDstCr = pMbCache->pSkipMb + 256 + 64;
const int32_t iOffsetY = (sCurMbMv[eSkipMode].iMvX >> 2) + (sCurMbMv[eSkipMode].iMvY >> 2) * iLineSizeY;
const int32_t iOffsetUV = (sCurMbMv[eSkipMode].iMvX >> 3) + (sCurMbMv[eSkipMode].iMvY >> 3) * iLineSizeUV;
if (!bQpSimilarFlag || !bMbSkipFlag) {
pDstLuma = pMbCache->pMemPredLuma;
pDstCb = pMbCache->pMemPredChroma;
pDstCr = pMbCache->pMemPredChroma + 64;
}
//MC
pFunc->sMcFuncs.pfLumaQuarpelMc[0] (pRefLuma + iOffsetY, iLineSizeY, pDstLuma, 16, 16);
pFunc->sMcFuncs.pfChromaMc (pRefCb + iOffsetUV, iLineSizeUV, pDstCb, 8, sMvp, 8, 8);
pFunc->sMcFuncs.pfChromaMc (pRefCr + iOffsetUV, iLineSizeUV, pDstCr, 8, sMvp, 8, 8);
pCurMb->uiCbp = 0;
pWelsMd->iCostLuma = 0;
pCurMb->pSadCost[0] = pFunc->sSampleDealingFuncs.pfSampleSad[BLOCK_16x16] (pMbCache->SPicData.pEncMb[0],
pCurDqLayer->iEncStride[0], pRefLuma + iOffsetY, iLineSizeY);
pWelsMd->iCostSkipMb = pCurMb->pSadCost[0];
ST16 (& (pCurMb->sP16x16Mv.iMvX), sCurMbMv[eSkipMode].iMvX);
ST16 (& (pCurMb->sP16x16Mv.iMvY), sCurMbMv[eSkipMode].iMvY);
ST16 (& (pCurDqLayer->pDecPic->sMvList[pCurMb->iMbXY].iMvX), sCurMbMv[eSkipMode].iMvX);
ST16 (& (pCurDqLayer->pDecPic->sMvList[pCurMb->iMbXY].iMvY), sCurMbMv[eSkipMode].iMvY);
if (bQpSimilarFlag && bMbSkipFlag) {
//update motion info to current MB
ST32 (pCurMb->pRefIndex, 0);
pFunc->pfUpdateMbMv (pCurMb->sMv, sMvp);
pCurMb->uiMbType = MB_TYPE_SKIP;
WelsRecPskip (pCurDqLayer, pEncCtx->pFuncList, pCurMb, pMbCache);
WelsMdInterUpdatePskip (pCurDqLayer, pSlice, pCurMb, pMbCache);
return;
}
pCurMb->uiMbType = MB_TYPE_16x16;
pWelsMd->sMe.sMe16x16.sMv.iMvX = sCurMbMv[eSkipMode].iMvX;
pWelsMd->sMe.sMe16x16.sMv.iMvY = sCurMbMv[eSkipMode].iMvY;
PredMv (&pMbCache->sMvComponents, 0, 4, 0, &pWelsMd->sMe.sMe16x16.sMvp);
pMbCache->sMbMvp[0] = pWelsMd->sMe.sMe16x16.sMvp;
UpdateP16x16MotionInfo (pMbCache, pCurMb, 0, &pWelsMd->sMe.sMe16x16.sMv);
if (pWelsMd->bMdUsingSad)
pWelsMd->iCostLuma = pCurMb->pSadCost[0];
else
pWelsMd->iCostLuma = pFunc->sSampleDealingFuncs.pfSampleSad[BLOCK_16x16] (pMbCache->SPicData.pEncMb[0],
pCurDqLayer->iEncStride[0], pRefLuma, iLineSizeY);
WelsInterMbEncode (pEncCtx, pSlice, pCurMb);
WelsPMbChromaEncode (pEncCtx, pSlice, pCurMb);
pFunc->pfCopy16x16Aligned (pMbCache->SPicData.pCsMb[0], pCurDqLayer->iCsStride[0], pMbCache->pMemPredLuma, 16);
pFunc->pfCopy8x8Aligned (pMbCache->SPicData.pCsMb[1], pCurDqLayer->iCsStride[1], pMbCache->pMemPredChroma, 8);
pFunc->pfCopy8x8Aligned (pMbCache->SPicData.pCsMb[2], pCurDqLayer->iCsStride[1], pMbCache->pMemPredChroma + 64, 8);
}
bool MdInterSCDPskipProcess (sWelsEncCtx* pEncCtx, SWelsMD* pWelsMd, SSlice* pSlice, SMB* pCurMb, SMbCache* pMbCache,
ESkipModes eSkipMode) {
SVAAFrameInfoExt_t* pVaaExt = static_cast<SVAAFrameInfoExt_t*> (pEncCtx->pVaa);
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
const int32_t kiRefMbQp = pCurDqLayer->pRefPic->pRefMbQp[pCurMb->iMbXY];
const int32_t kiCurMbQp = pCurMb->uiLumaQp;// unsigned -> signed
pJudgeSkipFun pJudeSkip[2] = {JudgeStaticSkip, JudgeScrollSkip};
bool bSkipFlag = pJudeSkip[eSkipMode] (pEncCtx, pCurMb, pMbCache, pWelsMd);
if (bSkipFlag) {
bool bQpSimilarFlag = (kiRefMbQp - kiCurMbQp <= DELTA_QP_SCD_THD || kiRefMbQp <= 26);
SMVUnitXY sVaaPredSkipMv = {0, 0}, sCurMbMv[2] = {{0, 0}, {0, 0}};
PredSkipMv (pMbCache, &sVaaPredSkipMv);
if (eSkipMode == SCROLLED) {
sCurMbMv[1].iMvX = static_cast<int16_t> (pVaaExt->sScrollDetectInfo.iScrollMvX << 2);
sCurMbMv[1].iMvY = static_cast<int16_t> (pVaaExt->sScrollDetectInfo.iScrollMvY << 2);
}
bool bMbSkipFlag = (LD32 (&sVaaPredSkipMv) == LD32 (&sCurMbMv[eSkipMode])) ;
SvcMdSCDMbEnc (pEncCtx, pWelsMd, pCurMb, pMbCache, pSlice, bQpSimilarFlag, bMbSkipFlag, sCurMbMv, eSkipMode);
return true;
}
return false;
}
void SetBlockStaticIdcToMd (void* pVaa, void* pMd, SMB* pCurMb, void* pDqLay) {
SVAAFrameInfoExt_t* pVaaExt = static_cast<SVAAFrameInfoExt_t*> (pVaa);
SWelsMD* pWelsMd = static_cast<SWelsMD*> (pMd);
SDqLayer* pDqLayer = static_cast<SDqLayer*> (pDqLay);
const int32_t kiMbX = pCurMb->iMbX;
const int32_t kiMbY = pCurMb->iMbY;
const int32_t kiMbWidth = pDqLayer->iMbWidth;
const int32_t kiWidth = kiMbWidth << 1;
const int32_t kiBlockIndexUp = (kiMbY << 1) * kiWidth + (kiMbX << 1);
const int32_t kiBlockIndexLow = ((kiMbY << 1) + 1) * kiWidth + (kiMbX << 1);
//fill_blockstaticidc with pVaaExt->pVaaBestBlockStaticIdc
pWelsMd->iBlock8x8StaticIdc[0] = pVaaExt->pVaaBestBlockStaticIdc[kiBlockIndexUp];
pWelsMd->iBlock8x8StaticIdc[1] = pVaaExt->pVaaBestBlockStaticIdc[kiBlockIndexUp + 1];
pWelsMd->iBlock8x8StaticIdc[2] = pVaaExt->pVaaBestBlockStaticIdc[kiBlockIndexLow];
pWelsMd->iBlock8x8StaticIdc[3] = pVaaExt->pVaaBestBlockStaticIdc[kiBlockIndexLow + 1];
}
///////////////////////
// Scene Change Detection (SCD) PSkip Decision for screen content
////////////////////////
bool WelsMdInterJudgeSCDPskip (void* pCtx, void* pMd, SSlice* slice, SMB* pCurMb, SMbCache* pMbCache) {
sWelsEncCtx* pEncCtx = (sWelsEncCtx*)pCtx;
SWelsMD* pWelsMd = (SWelsMD*)pMd;
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
SetBlockStaticIdcToMd (pEncCtx->pVaa, pWelsMd, pCurMb, pCurDqLayer);
//try static Pskip;
if (MdInterSCDPskipProcess (pEncCtx, pWelsMd, slice, pCurMb, pMbCache, STATIC)) {
return true;
}
//try scrolled Pskip
if (MdInterSCDPskipProcess (pEncCtx, pWelsMd, slice, pCurMb, pMbCache, SCROLLED)) {
return true;
}
return false;
}
bool WelsMdInterJudgeSCDPskipFalse (void* pEncCtx, void* pWelsMd, SSlice* slice, SMB* pCurMb, SMbCache* pMbCache) {
return false;
}
void WelsInitSCDPskipFunc (SWelsFuncPtrList* pFuncList, const bool bScrollingDetection) {
if (bScrollingDetection) {
pFuncList->pfSCDPSkipDecision = WelsMdInterJudgeSCDPskip;
} else {
pFuncList->pfSCDPSkipDecision = WelsMdInterJudgeSCDPskipFalse;
}
}
///////////////////////
// SubP16x16 Mode Decision for screen content
////////////////////////
//
//func pointer of inter MD for sub16x16 INTER MD for screen content coding
//
static inline void MergeSub16Me (const SWelsME& sSrcMe0, const SWelsME& sSrcMe1, SWelsME* pTarMe) {
memcpy (pTarMe, &sSrcMe0, sizeof (sSrcMe0)); // confirmed_safe_unsafe_usage
pTarMe->uiSadCost = sSrcMe0.uiSadCost + sSrcMe1.uiSadCost;//not precise cost since MVD cost is not the same
pTarMe->uiSatdCost = sSrcMe0.uiSatdCost + sSrcMe1.uiSatdCost;//not precise cost since MVD cost is not the same
}
static inline bool IsSameMv (const SMVUnitXY& sMv0, const SMVUnitXY& sMv1) {
return ((sMv0.iMvX == sMv1.iMvX) && (sMv0.iMvY == sMv1.iMvY));
}
bool TryModeMerge (SMbCache* pMbCache, SWelsMD* pWelsMd, SMB* pCurMb) {
SWelsME* pMe8x8 = & (pWelsMd->sMe.sMe8x8[0]);
const bool bSameMv16x8_0 = IsSameMv (pMe8x8[0].sMv, pMe8x8[1].sMv);
const bool bSameMv16x8_1 = IsSameMv (pMe8x8[2].sMv, pMe8x8[3].sMv);
const bool bSameMv8x16_0 = IsSameMv (pMe8x8[0].sMv, pMe8x8[2].sMv);
const bool bSameMv8x16_1 = IsSameMv (pMe8x8[1].sMv, pMe8x8[3].sMv);
//need to consider iRefIdx when multi ref is available
const bool bSameRefIdx16x8_0 = true; //pMe8x8[0].iRefIdx == pMe8x8[1].iRefIdx;
const bool bSameRefIdx16x8_1 = true; //pMe8x8[2].iRefIdx == pMe8x8[3].iRefIdx;
const bool bSameRefIdx8x16_0 = true; //pMe8x8[0].iRefIdx == pMe8x8[2].iRefIdx;
const bool bSameRefIdx8x16_1 = true; //pMe8x8[1].iRefIdx == pMe8x8[3].iRefIdx;
const int32_t iSameMv = ((bSameMv16x8_0 && bSameRefIdx16x8_0 && bSameMv16x8_1 && bSameRefIdx16x8_1) << 1) |
(bSameMv8x16_0 && bSameRefIdx8x16_0 && bSameMv8x16_1 && bSameRefIdx8x16_1);
//TODO: did not consider the MVD cost here, may consider later
switch (iSameMv) {
case 3:
//MERGE_16x16
//from test results of multiple sequences show that using the following 0x0F to merge 16x16
//for some seq there is BR saving some loss
//on the whole the BR will increase little bit
//to save complexity we decided not to merge 16x16 at present (10/12/2012)
//TODO: agjusted order, consider re-test later
break;
case 2:
pCurMb->uiMbType = MB_TYPE_16x8;
MergeSub16Me (pMe8x8[0], pMe8x8[1], & (pWelsMd->sMe.sMe16x8[0]));
MergeSub16Me (pMe8x8[2], pMe8x8[3], & (pWelsMd->sMe.sMe16x8[1]));
PredInter16x8Mv (pMbCache, 0, 0, & (pWelsMd->sMe.sMe16x8[0].sMvp));
PredInter16x8Mv (pMbCache, 8, 0, & (pWelsMd->sMe.sMe16x8[1].sMvp));
break;
case 1:
pCurMb->uiMbType = MB_TYPE_8x16;
MergeSub16Me (pMe8x8[0], pMe8x8[2], & (pWelsMd->sMe.sMe8x16[0]));
MergeSub16Me (pMe8x8[1], pMe8x8[3], & (pWelsMd->sMe.sMe8x16[1]));
PredInter8x16Mv (pMbCache, 0, 0, & (pWelsMd->sMe.sMe8x16[0].sMvp));
PredInter8x16Mv (pMbCache, 4, 0, & (pWelsMd->sMe.sMe8x16[1].sMvp));
break;
default:
break;
}
return (MB_TYPE_8x8 != pCurMb->uiMbType);
}
void WelsMdInterFinePartitionVaaOnScreen (void* pEnc, void* pMd, SSlice* pSlice, SMB* pCurMb, int32_t iBestCost) {
sWelsEncCtx* pEncCtx = (sWelsEncCtx*)pEnc;
SWelsMD* pWelsMd = (SWelsMD*)pMd;
SMbCache* pMbCache = &pSlice->sMbCacheInfo;
SDqLayer* pCurDqLayer = pEncCtx->pCurDqLayer;
int32_t iCostP8x8;
uint8_t uiMbSign = pEncCtx->pFuncList->pfGetMbSignFromInterVaa (&pEncCtx->pVaa->sVaaCalcInfo.pSad8x8[pCurMb->iMbXY][0]);
if (MBVAASIGN_FLAT == uiMbSign) {
return;
}
iCostP8x8 = WelsMdP8x8 (pEncCtx->pFuncList, pCurDqLayer, pWelsMd, pSlice);
if (iCostP8x8 < iBestCost) {
iBestCost = iCostP8x8;
pCurMb->uiMbType = MB_TYPE_8x8;
TryModeMerge (pMbCache, pWelsMd, pCurMb);
}
pWelsMd->iCostLuma = iBestCost;
}
//
// SetScrollingMvToMd
//
void SetScrollingMvToMd (void* pVaa, void* pMd) {
SVAAFrameInfoExt* pVaaExt = static_cast<SVAAFrameInfoExt*> (pVaa);
SWelsMD* pWelsMd = static_cast<SWelsMD*> (pMd);
SMVUnitXY sTempMv;
sTempMv.iMvX = pVaaExt->sScrollDetectInfo.iScrollMvX;
sTempMv.iMvY = pVaaExt->sScrollDetectInfo.iScrollMvY;
(pWelsMd->sMe.sMe16x16).sDirectionalMv =
(pWelsMd->sMe.sMe8x8[0]).sDirectionalMv =
(pWelsMd->sMe.sMe8x8[1]).sDirectionalMv =
(pWelsMd->sMe.sMe8x8[2]).sDirectionalMv =
(pWelsMd->sMe.sMe8x8[3]).sDirectionalMv = sTempMv;
}
void SetScrollingMvToMdNull (void* pVaa, void* pWelsMd) {
}
} // namespace WelsSVCEnc