ref: be9a1da2035540631d01cc40ac8243a84ae078ce
dir: /codec/decoder/core/src/mv_pred.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 mv_pred.c
*
* \brief Get MV predictor and update motion vector of mb cache
*
* \date 05/22/2009 Created
*
*************************************************************************************
*/
#include "mv_pred.h"
#include "ls_defines.h"
#include "mb_cache.h"
#include "parse_mb_syn_cabac.h"
namespace WelsDec {
static inline void SetRectBlock (void* vp, int32_t w, const int32_t h, int32_t stride, const uint32_t val,
const int32_t size) {
uint8_t* p = (uint8_t*)vp;
w *= size;
if (w == 1 && h == 4) {
* (uint8_t*) (p + 0 * stride) =
* (uint8_t*) (p + 1 * stride) =
* (uint8_t*) (p + 2 * stride) =
* (uint8_t*) (p + 3 * stride) = (uint8_t)val;
} else if (w == 2 && h == 2) {
* (uint16_t*) (p + 0 * stride) =
* (uint16_t*) (p + 1 * stride) = size == 4 ? (uint16_t)val : (uint16_t) (val * 0x0101U);
} else if (w == 2 && h == 4) {
* (uint16_t*) (p + 0 * stride) =
* (uint16_t*) (p + 1 * stride) =
* (uint16_t*) (p + 2 * stride) =
* (uint16_t*) (p + 3 * stride) = size == 4 ? (uint16_t)val : (uint16_t) (val * 0x0101U);
} else if (w == 4 && h == 2) {
* (uint32_t*) (p + 0 * stride) =
* (uint32_t*) (p + 1 * stride) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 4 && h == 4) {
* (uint32_t*) (p + 0 * stride) =
* (uint32_t*) (p + 1 * stride) =
* (uint32_t*) (p + 2 * stride) =
* (uint32_t*) (p + 3 * stride) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 8 && h == 1) {
* (uint32_t*) (p + 0 * stride) =
* (uint32_t*) (p + 0 * stride + 4) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 8 && h == 2) {
* (uint32_t*) (p + 0 * stride) =
* (uint32_t*) (p + 0 * stride + 4) =
* (uint32_t*) (p + 1 * stride) =
* (uint32_t*) (p + 1 * stride + 4) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 8 && h == 4) {
* (uint32_t*) (p + 0 * stride) =
* (uint32_t*) (p + 0 * stride + 4) =
* (uint32_t*) (p + 1 * stride) =
* (uint32_t*) (p + 1 * stride + 4) =
* (uint32_t*) (p + 2 * stride) =
* (uint32_t*) (p + 2 * stride + 4) =
* (uint32_t*) (p + 3 * stride) =
* (uint32_t*) (p + 3 * stride + 4) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 16 && h == 2) {
* (uint32_t*) (p + 0 * stride + 0) =
* (uint32_t*) (p + 0 * stride + 4) =
* (uint32_t*) (p + 0 * stride + 8) =
* (uint32_t*) (p + 0 * stride + 12) =
* (uint32_t*) (p + 1 * stride + 0) =
* (uint32_t*) (p + 1 * stride + 4) =
* (uint32_t*) (p + 1 * stride + 8) =
* (uint32_t*) (p + 1 * stride + 12) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 16 && h == 3) {
* (uint32_t*) (p + 0 * stride + 0) =
* (uint32_t*) (p + 0 * stride + 4) =
* (uint32_t*) (p + 0 * stride + 8) =
* (uint32_t*) (p + 0 * stride + 12) =
* (uint32_t*) (p + 1 * stride + 0) =
* (uint32_t*) (p + 1 * stride + 4) =
* (uint32_t*) (p + 1 * stride + 8) =
* (uint32_t*) (p + 1 * stride + 12) =
* (uint32_t*) (p + 2 * stride + 0) =
* (uint32_t*) (p + 2 * stride + 4) =
* (uint32_t*) (p + 2 * stride + 8) =
* (uint32_t*) (p + 2 * stride + 12) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
} else if (w == 16 && h == 4) {
* (uint32_t*) (p + 0 * stride + 0) =
* (uint32_t*) (p + 0 * stride + 4) =
* (uint32_t*) (p + 0 * stride + 8) =
* (uint32_t*) (p + 0 * stride + 12) =
* (uint32_t*) (p + 1 * stride + 0) =
* (uint32_t*) (p + 1 * stride + 4) =
* (uint32_t*) (p + 1 * stride + 8) =
* (uint32_t*) (p + 1 * stride + 12) =
* (uint32_t*) (p + 2 * stride + 0) =
* (uint32_t*) (p + 2 * stride + 4) =
* (uint32_t*) (p + 2 * stride + 8) =
* (uint32_t*) (p + 2 * stride + 12) =
* (uint32_t*) (p + 3 * stride + 0) =
* (uint32_t*) (p + 3 * stride + 4) =
* (uint32_t*) (p + 3 * stride + 8) =
* (uint32_t*) (p + 3 * stride + 12) = size == 4 ? val : (uint32_t) (val * 0x01010101UL);
}
}
void CopyRectBlock4Cols (void* vdst, void* vsrc, const int32_t stride_dst, const int32_t stride_src, int32_t w,
const int32_t size) {
uint8_t* dst = (uint8_t*)vdst;
uint8_t* src = (uint8_t*)vsrc;
w *= size;
if (w == 1) {
dst[stride_dst * 0] = src[stride_src * 0];
dst[stride_dst * 1] = src[stride_src * 1];
dst[stride_dst * 2] = src[stride_src * 2];
dst[stride_dst * 3] = src[stride_src * 3];
} else if (w == 2) {
* (uint16_t*) (&dst[stride_dst * 0]) = * (uint16_t*) (&src[stride_src * 0]);
* (uint16_t*) (&dst[stride_dst * 1]) = * (uint16_t*) (&src[stride_src * 1]);
* (uint16_t*) (&dst[stride_dst * 2]) = * (uint16_t*) (&src[stride_src * 2]);
* (uint16_t*) (&dst[stride_dst * 3]) = * (uint16_t*) (&src[stride_src * 3]);
} else if (w == 4) {
* (uint32_t*) (&dst[stride_dst * 0]) = * (uint32_t*) (&src[stride_src * 0]);
* (uint32_t*) (&dst[stride_dst * 1]) = * (uint32_t*) (&src[stride_src * 1]);
* (uint32_t*) (&dst[stride_dst * 2]) = * (uint32_t*) (&src[stride_src * 2]);
* (uint32_t*) (&dst[stride_dst * 3]) = * (uint32_t*) (&src[stride_src * 3]);
} else if (w == 16) {
memcpy (&dst[stride_dst * 0], &src[stride_src * 0], 16);
memcpy (&dst[stride_dst * 1], &src[stride_src * 1], 16);
memcpy (&dst[stride_dst * 2], &src[stride_src * 2], 16);
memcpy (&dst[stride_dst * 3], &src[stride_src * 3], 16);
}
}
void PredPSkipMvFromNeighbor (PDqLayer pCurDqLayer, int16_t iMvp[2]) {
bool bTopAvail, bLeftTopAvail, bRightTopAvail, bLeftAvail;
int32_t iCurSliceIdc, iTopSliceIdc, iLeftTopSliceIdc, iRightTopSliceIdc, iLeftSliceIdc;
int32_t iLeftTopType, iRightTopType, iTopType, iLeftType;
int32_t iCurX, iCurY, iCurXy, iLeftXy, iTopXy = 0, iLeftTopXy = 0, iRightTopXy = 0;
int8_t iLeftRef;
int8_t iTopRef;
int8_t iRightTopRef;
int8_t iLeftTopRef;
int8_t iDiagonalRef;
int8_t iMatchRef;
int16_t iMvA[2], iMvB[2], iMvC[2], iMvD[2];
iCurXy = pCurDqLayer->iMbXyIndex;
iCurX = pCurDqLayer->iMbX;
iCurY = pCurDqLayer->iMbY;
iCurSliceIdc = pCurDqLayer->pSliceIdc[iCurXy];
if (iCurX != 0) {
iLeftXy = iCurXy - 1;
iLeftSliceIdc = pCurDqLayer->pSliceIdc[iLeftXy];
bLeftAvail = (iLeftSliceIdc == iCurSliceIdc);
} else {
bLeftAvail = 0;
bLeftTopAvail = 0;
}
if (iCurY != 0) {
iTopXy = iCurXy - pCurDqLayer->iMbWidth;
iTopSliceIdc = pCurDqLayer->pSliceIdc[iTopXy];
bTopAvail = (iTopSliceIdc == iCurSliceIdc);
if (iCurX != 0) {
iLeftTopXy = iTopXy - 1;
iLeftTopSliceIdc = pCurDqLayer->pSliceIdc[iLeftTopXy];
bLeftTopAvail = (iLeftTopSliceIdc == iCurSliceIdc);
} else {
bLeftTopAvail = 0;
}
if (iCurX != (pCurDqLayer->iMbWidth - 1)) {
iRightTopXy = iTopXy + 1;
iRightTopSliceIdc = pCurDqLayer->pSliceIdc[iRightTopXy];
bRightTopAvail = (iRightTopSliceIdc == iCurSliceIdc);
} else {
bRightTopAvail = 0;
}
} else {
bTopAvail = 0;
bLeftTopAvail = 0;
bRightTopAvail = 0;
}
iLeftType = ((iCurX != 0 && bLeftAvail) ? GetMbType (pCurDqLayer)[iLeftXy] : 0);
iTopType = ((iCurY != 0 && bTopAvail) ? GetMbType (pCurDqLayer)[iTopXy] : 0);
iLeftTopType = ((iCurX != 0 && iCurY != 0 && bLeftTopAvail)
? GetMbType (pCurDqLayer)[iLeftTopXy] : 0);
iRightTopType = ((iCurX != pCurDqLayer->iMbWidth - 1 && iCurY != 0 && bRightTopAvail)
? GetMbType (pCurDqLayer)[iRightTopXy] : 0);
/*get neb mv&iRefIdxArray*/
/*left*/
if (bLeftAvail && IS_INTER (iLeftType)) {
ST32 (iMvA, LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[0][iLeftXy][3] : pCurDqLayer->pMv[0][iLeftXy][3]));
iLeftRef = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[0][iLeftXy][3] : pCurDqLayer->pRefIndex[0][iLeftXy][3];
} else {
ST32 (iMvA, 0);
if (0 == bLeftAvail) { //not available
iLeftRef = REF_NOT_AVAIL;
} else { //available but is intra mb type
iLeftRef = REF_NOT_IN_LIST;
}
}
if (REF_NOT_AVAIL == iLeftRef ||
(0 == iLeftRef && 0 == * (int32_t*)iMvA)) {
ST32 (iMvp, 0);
return;
}
/*top*/
if (bTopAvail && IS_INTER (iTopType)) {
ST32 (iMvB, LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[0][iTopXy][12] : pCurDqLayer->pMv[0][iTopXy][12]));
iTopRef = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[0][iTopXy][12] : pCurDqLayer->pRefIndex[0][iTopXy][12];
} else {
ST32 (iMvB, 0);
if (0 == bTopAvail) { //not available
iTopRef = REF_NOT_AVAIL;
} else { //available but is intra mb type
iTopRef = REF_NOT_IN_LIST;
}
}
if (REF_NOT_AVAIL == iTopRef ||
(0 == iTopRef && 0 == * (int32_t*)iMvB)) {
ST32 (iMvp, 0);
return;
}
/*right_top*/
if (bRightTopAvail && IS_INTER (iRightTopType)) {
ST32 (iMvC, LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[0][iRightTopXy][12] :
pCurDqLayer->pMv[0][iRightTopXy][12]));
iRightTopRef = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[0][iRightTopXy][12] :
pCurDqLayer->pRefIndex[0][iRightTopXy][12];
} else {
ST32 (iMvC, 0);
if (0 == bRightTopAvail) { //not available
iRightTopRef = REF_NOT_AVAIL;
} else { //available but is intra mb type
iRightTopRef = REF_NOT_IN_LIST;
}
}
/*left_top*/
if (bLeftTopAvail && IS_INTER (iLeftTopType)) {
ST32 (iMvD, LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[0][iLeftTopXy][15] : pCurDqLayer->pMv[0][iLeftTopXy][15]));
iLeftTopRef = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[0][iLeftTopXy][15] :
pCurDqLayer->pRefIndex[0][iLeftTopXy][15];
} else {
ST32 (iMvD, 0);
if (0 == bLeftTopAvail) { //not available
iLeftTopRef = REF_NOT_AVAIL;
} else { //available but is intra mb type
iLeftTopRef = REF_NOT_IN_LIST;
}
}
iDiagonalRef = iRightTopRef;
if (REF_NOT_AVAIL == iDiagonalRef) {
iDiagonalRef = iLeftTopRef;
* (int32_t*)iMvC = * (int32_t*)iMvD;
}
if (REF_NOT_AVAIL == iTopRef && REF_NOT_AVAIL == iDiagonalRef && iLeftRef >= REF_NOT_IN_LIST) {
ST32 (iMvp, LD32 (iMvA));
return;
}
iMatchRef = (0 == iLeftRef) + (0 == iTopRef) + (0 == iDiagonalRef);
if (1 == iMatchRef) {
if (0 == iLeftRef) {
ST32 (iMvp, LD32 (iMvA));
} else if (0 == iTopRef) {
ST32 (iMvp, LD32 (iMvB));
} else {
ST32 (iMvp, LD32 (iMvC));
}
} else {
iMvp[0] = WelsMedian (iMvA[0], iMvB[0], iMvC[0]);
iMvp[1] = WelsMedian (iMvA[1], iMvB[1], iMvC[1]);
}
}
int32_t GetColocatedMb (PWelsDecoderContext pCtx, MbType& mbType, SubMbType& subMbType) {
PDqLayer pCurDqLayer = pCtx->pCurDqLayer;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
uint32_t is8x8 = IS_Inter_8x8 (GetMbType (pCurDqLayer)[iMbXy]);
mbType = GetMbType (pCurDqLayer)[iMbXy];
PPicture colocPic = pCtx->sRefPic.pRefList[LIST_1][0];
if (GetThreadCount (pCtx) > 1) {
if (16 * pCurDqLayer->iMbY > pCtx->lastReadyHeightOffset[1][0]) {
if (colocPic->pReadyEvent[pCurDqLayer->iMbY].isSignaled != 1) {
WAIT_EVENT (&colocPic->pReadyEvent[pCurDqLayer->iMbY], WELS_DEC_THREAD_WAIT_INFINITE);
}
pCtx->lastReadyHeightOffset[1][0] = 16 * pCurDqLayer->iMbY;
}
}
if (colocPic == NULL) {
SLogContext* pLogCtx = & (pCtx->sLogCtx);
WelsLog (pLogCtx, WELS_LOG_ERROR, "Colocated Ref Picture for B-Slice is lost, B-Slice decoding cannot be continued!");
return GENERATE_ERROR_NO (ERR_LEVEL_SLICE_DATA, ERR_INFO_REFERENCE_PIC_LOST);
}
MbType coloc_mbType = colocPic->pMbType[iMbXy];
if (coloc_mbType == MB_TYPE_SKIP) {
//This indicates the colocated MB is P SKIP MB
coloc_mbType |= MB_TYPE_16x16 | MB_TYPE_P0L0 | MB_TYPE_P1L0;
}
if (IS_Inter_8x8 (coloc_mbType) && !pCtx->pSps->bDirect8x8InferenceFlag) {
subMbType = SUB_MB_TYPE_4x4 | MB_TYPE_P0L0 | MB_TYPE_P0L1 | MB_TYPE_DIRECT;
mbType |= MB_TYPE_8x8 | MB_TYPE_L0 | MB_TYPE_L1;
} else if (!is8x8 && (IS_INTER_16x16 (coloc_mbType) || IS_INTRA (coloc_mbType)/* || IS_SKIP(coloc_mbType)*/)) {
subMbType = SUB_MB_TYPE_8x8 | MB_TYPE_P0L0 | MB_TYPE_P0L1 | MB_TYPE_DIRECT;
mbType |= MB_TYPE_16x16 | MB_TYPE_L0 | MB_TYPE_L1;
} else {
subMbType = SUB_MB_TYPE_8x8 | MB_TYPE_P0L0 | MB_TYPE_P0L1 | MB_TYPE_DIRECT;
mbType |= MB_TYPE_8x8 | MB_TYPE_L0 | MB_TYPE_L1;
}
if (IS_INTRA (coloc_mbType)) {
SetRectBlock (pCurDqLayer->iColocIntra, 4, 4, 4 * sizeof (int8_t), 1, sizeof (int8_t));
return ERR_NONE;
}
SetRectBlock (pCurDqLayer->iColocIntra, 4, 4, 4 * sizeof (int8_t), 0, sizeof (int8_t));
if (IS_INTER_16x16 (mbType)) {
int16_t iMVZero[2] = { 0 };
int16_t* pMv = IS_TYPE_L1 (coloc_mbType) ? colocPic->pMv[LIST_1][iMbXy][0] : iMVZero;
ST32 (pCurDqLayer->iColocMv[LIST_0][0], LD32 (colocPic->pMv[LIST_0][iMbXy][0]));
ST32 (pCurDqLayer->iColocMv[LIST_1][0], LD32 (pMv));
pCurDqLayer->iColocRefIndex[LIST_0][0] = colocPic->pRefIndex[LIST_0][iMbXy][0];
pCurDqLayer->iColocRefIndex[LIST_1][0] = IS_TYPE_L1 (coloc_mbType) ? colocPic->pRefIndex[LIST_1][iMbXy][0] :
REF_NOT_IN_LIST;
} else {
if (!pCtx->pSps->bDirect8x8InferenceFlag) {
CopyRectBlock4Cols (pCurDqLayer->iColocMv[LIST_0], colocPic->pMv[LIST_0][iMbXy], 16, 16, 4, 4);
CopyRectBlock4Cols (pCurDqLayer->iColocRefIndex[LIST_0], colocPic->pRefIndex[LIST_0][iMbXy], 4, 4, 4, 1);
if (IS_TYPE_L1 (coloc_mbType)) {
CopyRectBlock4Cols (pCurDqLayer->iColocMv[LIST_1], colocPic->pMv[LIST_1][iMbXy], 16, 16, 4, 4);
CopyRectBlock4Cols (pCurDqLayer->iColocRefIndex[LIST_1], colocPic->pRefIndex[LIST_1][iMbXy], 4, 4, 4, 1);
} else { // only forward prediction
SetRectBlock (pCurDqLayer->iColocRefIndex[LIST_1], 4, 4, 4, (uint8_t)REF_NOT_IN_LIST, 1);
}
} else {
for (int32_t listIdx = 0; listIdx < 1 + !! (coloc_mbType & MB_TYPE_L1); listIdx++) {
SetRectBlock (pCurDqLayer->iColocMv[listIdx][0], 2, 2, 16, LD32 (colocPic->pMv[listIdx][iMbXy][0]), 4);
SetRectBlock (pCurDqLayer->iColocMv[listIdx][2], 2, 2, 16, LD32 (colocPic->pMv[listIdx][iMbXy][3]), 4);
SetRectBlock (pCurDqLayer->iColocMv[listIdx][8], 2, 2, 16, LD32 (colocPic->pMv[listIdx][iMbXy][12]), 4);
SetRectBlock (pCurDqLayer->iColocMv[listIdx][10], 2, 2, 16, LD32 (colocPic->pMv[listIdx][iMbXy][15]), 4);
SetRectBlock (&pCurDqLayer->iColocRefIndex[listIdx][0], 2, 2, 4, colocPic->pRefIndex[listIdx][iMbXy][0], 1);
SetRectBlock (&pCurDqLayer->iColocRefIndex[listIdx][2], 2, 2, 4, colocPic->pRefIndex[listIdx][iMbXy][3], 1);
SetRectBlock (&pCurDqLayer->iColocRefIndex[listIdx][8], 2, 2, 4, colocPic->pRefIndex[listIdx][iMbXy][12], 1);
SetRectBlock (&pCurDqLayer->iColocRefIndex[listIdx][10], 2, 2, 4, colocPic->pRefIndex[listIdx][iMbXy][15], 1);
}
if (! (coloc_mbType & MB_TYPE_L1)) // only forward prediction
SetRectBlock (&pCurDqLayer->iColocRefIndex[1][0], 4, 4, 4, (uint8_t)REF_NOT_IN_LIST, 1);
}
}
return ERR_NONE;
}
int32_t PredMvBDirectSpatial (PWelsDecoderContext pCtx, int16_t iMvp[LIST_A][2], int8_t ref[LIST_A],
SubMbType& subMbType) {
int32_t ret = ERR_NONE;
PDqLayer pCurDqLayer = pCtx->pCurDqLayer;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
bool bSkipOrDirect = (IS_SKIP (GetMbType (pCurDqLayer)[iMbXy]) | IS_DIRECT (GetMbType (pCurDqLayer)[iMbXy])) > 0;
MbType mbType;
ret = GetColocatedMb (pCtx, mbType, subMbType);
if (ret != ERR_NONE) {
return ret;
}
bool bTopAvail, bLeftTopAvail, bRightTopAvail, bLeftAvail;
int32_t iLeftTopType, iRightTopType, iTopType, iLeftType;
int32_t iCurSliceIdc, iTopSliceIdc, iLeftTopSliceIdc, iRightTopSliceIdc, iLeftSliceIdc;
int32_t iCurX, iCurY, iCurXy, iLeftXy = 0, iTopXy = 0, iLeftTopXy = 0, iRightTopXy = 0;
int8_t iLeftRef[LIST_A];
int8_t iTopRef[LIST_A];
int8_t iRightTopRef[LIST_A];
int8_t iLeftTopRef[LIST_A];
int8_t iDiagonalRef[LIST_A];
int16_t iMvA[LIST_A][2], iMvB[LIST_A][2], iMvC[LIST_A][2], iMvD[LIST_A][2];
iCurXy = pCurDqLayer->iMbXyIndex;
iCurX = pCurDqLayer->iMbX;
iCurY = pCurDqLayer->iMbY;
iCurSliceIdc = pCurDqLayer->pSliceIdc[iCurXy];
if (iCurX != 0) {
iLeftXy = iCurXy - 1;
iLeftSliceIdc = pCurDqLayer->pSliceIdc[iLeftXy];
bLeftAvail = (iLeftSliceIdc == iCurSliceIdc);
} else {
bLeftAvail = 0;
bLeftTopAvail = 0;
}
if (iCurY != 0) {
iTopXy = iCurXy - pCurDqLayer->iMbWidth;
iTopSliceIdc = pCurDqLayer->pSliceIdc[iTopXy];
bTopAvail = (iTopSliceIdc == iCurSliceIdc);
if (iCurX != 0) {
iLeftTopXy = iTopXy - 1;
iLeftTopSliceIdc = pCurDqLayer->pSliceIdc[iLeftTopXy];
bLeftTopAvail = (iLeftTopSliceIdc == iCurSliceIdc);
} else {
bLeftTopAvail = 0;
}
if (iCurX != (pCurDqLayer->iMbWidth - 1)) {
iRightTopXy = iTopXy + 1;
iRightTopSliceIdc = pCurDqLayer->pSliceIdc[iRightTopXy];
bRightTopAvail = (iRightTopSliceIdc == iCurSliceIdc);
} else {
bRightTopAvail = 0;
}
} else {
bTopAvail = 0;
bLeftTopAvail = 0;
bRightTopAvail = 0;
}
iLeftType = ((iCurX != 0 && bLeftAvail) ? GetMbType (pCurDqLayer)[iLeftXy] : 0);
iTopType = ((iCurY != 0 && bTopAvail) ? GetMbType (pCurDqLayer)[iTopXy] : 0);
iLeftTopType = ((iCurX != 0 && iCurY != 0 && bLeftTopAvail)
? GetMbType (pCurDqLayer)[iLeftTopXy] : 0);
iRightTopType = ((iCurX != pCurDqLayer->iMbWidth - 1 && iCurY != 0 && bRightTopAvail)
? GetMbType (pCurDqLayer)[iRightTopXy] : 0);
/*get neb mv&iRefIdxArray*/
for (int32_t listIdx = LIST_0; listIdx < LIST_A; ++listIdx) {
/*left*/
if (bLeftAvail && IS_INTER (iLeftType)) {
ST32 (iMvA[listIdx], LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[listIdx][iLeftXy][3] :
pCurDqLayer->pMv[listIdx][iLeftXy][3]));
iLeftRef[listIdx] = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[listIdx][iLeftXy][3] :
pCurDqLayer->pRefIndex[listIdx][iLeftXy][3];
} else {
ST32 (iMvA[listIdx], 0);
if (0 == bLeftAvail) { //not available
iLeftRef[listIdx] = REF_NOT_AVAIL;
} else { //available but is intra mb type
iLeftRef[listIdx] = REF_NOT_IN_LIST;
}
}
/*top*/
if (bTopAvail && IS_INTER (iTopType)) {
ST32 (iMvB[listIdx], LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[listIdx][iTopXy][12] :
pCurDqLayer->pMv[listIdx][iTopXy][12]));
iTopRef[listIdx] = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[listIdx][iTopXy][12] :
pCurDqLayer->pRefIndex[listIdx][iTopXy][12];
} else {
ST32 (iMvB[listIdx], 0);
if (0 == bTopAvail) { //not available
iTopRef[listIdx] = REF_NOT_AVAIL;
} else { //available but is intra mb type
iTopRef[listIdx] = REF_NOT_IN_LIST;
}
}
/*right_top*/
if (bRightTopAvail && IS_INTER (iRightTopType)) {
ST32 (iMvC[listIdx], LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[listIdx][iRightTopXy][12] :
pCurDqLayer->pMv[listIdx][iRightTopXy][12]));
iRightTopRef[listIdx] = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[listIdx][iRightTopXy][12] :
pCurDqLayer->pRefIndex[listIdx][iRightTopXy][12];
} else {
ST32 (iMvC[listIdx], 0);
if (0 == bRightTopAvail) { //not available
iRightTopRef[listIdx] = REF_NOT_AVAIL;
} else { //available but is intra mb type
iRightTopRef[listIdx] = REF_NOT_IN_LIST;
}
}
/*left_top*/
if (bLeftTopAvail && IS_INTER (iLeftTopType)) {
ST32 (iMvD[listIdx], LD32 (pCurDqLayer->pDec ? pCurDqLayer->pDec->pMv[listIdx][iLeftTopXy][15] :
pCurDqLayer->pMv[listIdx][iLeftTopXy][15]));
iLeftTopRef[listIdx] = pCurDqLayer->pDec ? pCurDqLayer->pDec->pRefIndex[listIdx][iLeftTopXy][15] :
pCurDqLayer->pRefIndex[listIdx][iLeftTopXy][15];
} else {
ST32 (iMvD[listIdx], 0);
if (0 == bLeftTopAvail) { //not available
iLeftTopRef[listIdx] = REF_NOT_AVAIL;
} else { //available but is intra mb type
iLeftTopRef[listIdx] = REF_NOT_IN_LIST;
}
}
iDiagonalRef[listIdx] = iRightTopRef[listIdx];
if (REF_NOT_AVAIL == iDiagonalRef[listIdx]) {
iDiagonalRef[listIdx] = iLeftTopRef[listIdx];
ST32 (iMvC[listIdx], LD32 (iMvD[listIdx]));
}
int8_t ref_temp = WELS_MIN_POSITIVE (iTopRef[listIdx], iDiagonalRef[listIdx]);
ref[listIdx] = WELS_MIN_POSITIVE (iLeftRef[listIdx], ref_temp);
if (ref[listIdx] >= 0) {
uint32_t match_count = (iLeftRef[listIdx] == ref[listIdx]) + (iTopRef[listIdx] == ref[listIdx]) +
(iDiagonalRef[listIdx] == ref[listIdx]);
if (match_count == 1) {
if (iLeftRef[listIdx] == ref[listIdx]) {
ST32 (iMvp[listIdx], LD32 (iMvA[listIdx]));
} else if (iTopRef[listIdx] == ref[listIdx]) {
ST32 (iMvp[listIdx], LD32 (iMvB[listIdx]));
} else {
ST32 (iMvp[listIdx], LD32 (iMvC[listIdx]));
}
} else {
iMvp[listIdx][0] = WelsMedian (iMvA[listIdx][0], iMvB[listIdx][0], iMvC[listIdx][0]);
iMvp[listIdx][1] = WelsMedian (iMvA[listIdx][1], iMvB[listIdx][1], iMvC[listIdx][1]);
}
} else {
iMvp[listIdx][0] = 0;
iMvp[listIdx][1] = 0;
ref[listIdx] = REF_NOT_IN_LIST;
}
}
if (ref[LIST_0] <= REF_NOT_IN_LIST && ref[LIST_1] <= REF_NOT_IN_LIST) {
ref[LIST_0] = ref[LIST_1] = 0;
} else if (ref[LIST_1] < 0) {
mbType &= ~MB_TYPE_L1;
subMbType &= ~MB_TYPE_L1;
} else if (ref[LIST_0] < 0) {
mbType &= ~MB_TYPE_L0;
subMbType &= ~MB_TYPE_L0;
}
GetMbType (pCurDqLayer)[iMbXy] = mbType;
int16_t pMvd[4] = { 0 };
bool bIsLongRef = pCtx->sRefPic.pRefList[LIST_1][0]->bIsLongRef;
if (IS_INTER_16x16 (mbType)) {
if ((* (int32_t*)iMvp[LIST_0] | * (int32_t*)iMvp[LIST_1])) {
if (0 == pCurDqLayer->iColocIntra[0] && !bIsLongRef
&& ((pCurDqLayer->iColocRefIndex[LIST_0][0] == 0 && (unsigned) (pCurDqLayer->iColocMv[LIST_0][0][0] + 1) <= 2
&& (unsigned) (pCurDqLayer->iColocMv[LIST_0][0][1] + 1) <= 2)
|| (pCurDqLayer->iColocRefIndex[LIST_0][0] < 0 && pCurDqLayer->iColocRefIndex[LIST_1][0] == 0
&& (unsigned) (pCurDqLayer->iColocMv[LIST_1][0][0] + 1) <= 2
&& (unsigned) (pCurDqLayer->iColocMv[LIST_1][0][1] + 1) <= 2))) {
if (0 >= ref[0]) * (uint32_t*)iMvp[LIST_0] = 0;
if (0 >= ref[1]) * (uint32_t*)iMvp[LIST_1] = 0;
}
}
UpdateP16x16DirectCabac (pCurDqLayer);
for (int32_t listIdx = LIST_0; listIdx < LIST_A; ++listIdx) {
UpdateP16x16MotionInfo (pCurDqLayer, listIdx, ref[listIdx], iMvp[listIdx]);
UpdateP16x16MvdCabac (pCurDqLayer, pMvd, listIdx);
}
} else {
if (bSkipOrDirect) {
int8_t pSubPartCount[4], pPartW[4];
for (int32_t i = 0; i < 4; i++) { //Direct 8x8 Ref and mv
int16_t iIdx8 = i << 2;
pCurDqLayer->pSubMbType[iMbXy][i] = subMbType;
int8_t pRefIndex[LIST_A][30];
UpdateP8x8RefIdxCabac (pCurDqLayer, pRefIndex, iIdx8, ref[LIST_0], LIST_0);
UpdateP8x8RefIdxCabac (pCurDqLayer, pRefIndex, iIdx8, ref[LIST_1], LIST_1);
UpdateP8x8DirectCabac (pCurDqLayer, iIdx8);
pSubPartCount[i] = g_ksInterBSubMbTypeInfo[0].iPartCount;
pPartW[i] = g_ksInterBSubMbTypeInfo[0].iPartWidth;
if (IS_SUB_4x4 (subMbType)) {
pSubPartCount[i] = 4;
pPartW[i] = 1;
}
FillSpatialDirect8x8Mv (pCurDqLayer, iIdx8, pSubPartCount[i], pPartW[i], subMbType, bIsLongRef, iMvp, ref, NULL, NULL);
}
}
}
return ret;
}
int32_t PredBDirectTemporal (PWelsDecoderContext pCtx, int16_t iMvp[LIST_A][2], int8_t ref[LIST_A],
SubMbType& subMbType) {
int32_t ret = ERR_NONE;
PDqLayer pCurDqLayer = pCtx->pCurDqLayer;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
bool bSkipOrDirect = (IS_SKIP (GetMbType (pCurDqLayer)[iMbXy]) | IS_DIRECT (GetMbType (pCurDqLayer)[iMbXy])) > 0;
MbType mbType;
ret = GetColocatedMb (pCtx, mbType, subMbType);
if (ret != ERR_NONE) {
return ret;
}
GetMbType (pCurDqLayer)[iMbXy] = mbType;
PSlice pSlice = &pCurDqLayer->sLayerInfo.sSliceInLayer;
PSliceHeader pSliceHeader = &pSlice->sSliceHeaderExt.sSliceHeader;
int16_t pMvd[4] = { 0 };
const int32_t ref0Count = WELS_MIN (pSliceHeader->uiRefCount[LIST_0], pCtx->sRefPic.uiRefCount[LIST_0]);
if (IS_INTER_16x16 (mbType)) {
ref[LIST_0] = 0;
ref[LIST_1] = 0;
UpdateP16x16DirectCabac (pCurDqLayer);
UpdateP16x16RefIdx (pCurDqLayer, LIST_1, ref[LIST_1]);
ST64 (iMvp, 0);
if (pCurDqLayer->iColocIntra[0]) {
UpdateP16x16MotionOnly (pCurDqLayer, LIST_0, iMvp[LIST_0]);
UpdateP16x16MotionOnly (pCurDqLayer, LIST_1, iMvp[LIST_1]);
UpdateP16x16RefIdx (pCurDqLayer, LIST_0, ref[LIST_0]);
} else {
ref[LIST_0] = 0;
int16_t* mv = pCurDqLayer->iColocMv[LIST_0][0];
int8_t colocRefIndexL0 = pCurDqLayer->iColocRefIndex[LIST_0][0];
if (colocRefIndexL0 >= 0) {
ref[LIST_0] = MapColToList0 (pCtx, colocRefIndexL0, ref0Count);
} else {
mv = pCurDqLayer->iColocMv[LIST_1][0];
}
UpdateP16x16RefIdx (pCurDqLayer, LIST_0, ref[LIST_0]);
iMvp[LIST_0][0] = (pSlice->iMvScale[LIST_0][ref[LIST_0]] * mv[0] + 128) >> 8;
iMvp[LIST_0][1] = (pSlice->iMvScale[LIST_0][ref[LIST_0]] * mv[1] + 128) >> 8;
UpdateP16x16MotionOnly (pCurDqLayer, LIST_0, iMvp[LIST_0]);
iMvp[LIST_1][0] = iMvp[LIST_0][0] - mv[0];
iMvp[LIST_1][1] = iMvp[LIST_0][1] - mv[1];
UpdateP16x16MotionOnly (pCurDqLayer, LIST_1, iMvp[LIST_1]);
}
UpdateP16x16MvdCabac (pCurDqLayer, pMvd, LIST_0);
UpdateP16x16MvdCabac (pCurDqLayer, pMvd, LIST_1);
} else {
if (bSkipOrDirect) {
int8_t pSubPartCount[4], pPartW[4];
int8_t pRefIndex[LIST_A][30];
for (int32_t i = 0; i < 4; i++) {
int16_t iIdx8 = i << 2;
const uint8_t iScan4Idx = g_kuiScan4[iIdx8];
pCurDqLayer->pSubMbType[iMbXy][i] = subMbType;
int16_t (*mvColoc)[2] = pCurDqLayer->iColocMv[LIST_0];
ref[LIST_1] = 0;
UpdateP8x8RefIdxCabac (pCurDqLayer, pRefIndex, iIdx8, ref[LIST_1], LIST_1);
if (pCurDqLayer->iColocIntra[iScan4Idx]) {
ref[LIST_0] = 0;
UpdateP8x8RefIdxCabac (pCurDqLayer, pRefIndex, iIdx8, ref[LIST_0], LIST_0);
ST64 (iMvp, 0);
} else {
ref[LIST_0] = 0;
int8_t colocRefIndexL0 = pCurDqLayer->iColocRefIndex[LIST_0][iScan4Idx];
if (colocRefIndexL0 >= 0) {
ref[LIST_0] = MapColToList0 (pCtx, colocRefIndexL0, ref0Count);
} else {
mvColoc = pCurDqLayer->iColocMv[LIST_1];
}
UpdateP8x8RefIdxCabac (pCurDqLayer, pRefIndex, iIdx8, ref[LIST_0], LIST_0);
}
UpdateP8x8DirectCabac (pCurDqLayer, iIdx8);
pSubPartCount[i] = g_ksInterBSubMbTypeInfo[0].iPartCount;
pPartW[i] = g_ksInterBSubMbTypeInfo[0].iPartWidth;
if (IS_SUB_4x4 (subMbType)) {
pSubPartCount[i] = 4;
pPartW[i] = 1;
}
FillTemporalDirect8x8Mv (pCurDqLayer, iIdx8, pSubPartCount[i], pPartW[i], subMbType, ref, mvColoc, NULL, NULL);
}
}
}
return ret;
}
//basic iMVs prediction unit for iMVs partition width (4, 2, 1)
void PredMv (int16_t iMotionVector[LIST_A][30][MV_A], int8_t iRefIndex[LIST_A][30],
int32_t listIdx, int32_t iPartIdx, int32_t iPartWidth, int8_t iRef, int16_t iMVP[2]) {
const uint8_t kuiLeftIdx = g_kuiCache30ScanIdx[iPartIdx] - 1;
const uint8_t kuiTopIdx = g_kuiCache30ScanIdx[iPartIdx] - 6;
const uint8_t kuiRightTopIdx = kuiTopIdx + iPartWidth;
const uint8_t kuiLeftTopIdx = kuiTopIdx - 1;
const int8_t kiLeftRef = iRefIndex[listIdx][kuiLeftIdx];
const int8_t kiTopRef = iRefIndex[listIdx][ kuiTopIdx];
const int8_t kiRightTopRef = iRefIndex[listIdx][kuiRightTopIdx];
const int8_t kiLeftTopRef = iRefIndex[listIdx][ kuiLeftTopIdx];
int8_t iDiagonalRef = kiRightTopRef;
int8_t iMatchRef = 0;
int16_t iAMV[2], iBMV[2], iCMV[2];
ST32 (iAMV, LD32 (iMotionVector[listIdx][ kuiLeftIdx]));
ST32 (iBMV, LD32 (iMotionVector[listIdx][ kuiTopIdx]));
ST32 (iCMV, LD32 (iMotionVector[listIdx][kuiRightTopIdx]));
if (REF_NOT_AVAIL == iDiagonalRef) {
iDiagonalRef = kiLeftTopRef;
ST32 (iCMV, LD32 (iMotionVector[listIdx][kuiLeftTopIdx]));
}
iMatchRef = (iRef == kiLeftRef) + (iRef == kiTopRef) + (iRef == iDiagonalRef);
if (REF_NOT_AVAIL == kiTopRef && REF_NOT_AVAIL == iDiagonalRef && kiLeftRef >= REF_NOT_IN_LIST) {
ST32 (iMVP, LD32 (iAMV));
return;
}
if (1 == iMatchRef) {
if (iRef == kiLeftRef) {
ST32 (iMVP, LD32 (iAMV));
} else if (iRef == kiTopRef) {
ST32 (iMVP, LD32 (iBMV));
} else {
ST32 (iMVP, LD32 (iCMV));
}
} else {
iMVP[0] = WelsMedian (iAMV[0], iBMV[0], iCMV[0]);
iMVP[1] = WelsMedian (iAMV[1], iBMV[1], iCMV[1]);
}
}
void PredInter8x16Mv (int16_t iMotionVector[LIST_A][30][MV_A], int8_t iRefIndex[LIST_A][30],
int32_t listIdx, int32_t iPartIdx, int8_t iRef, int16_t iMVP[2]) {
if (0 == iPartIdx) {
const int8_t kiLeftRef = iRefIndex[listIdx][6];
if (iRef == kiLeftRef) {
ST32 (iMVP, LD32 (&iMotionVector[listIdx][6][0]));
return;
}
} else { // 1 == iPartIdx
int8_t iDiagonalRef = iRefIndex[listIdx][5]; //top-right
int8_t index = 5;
if (REF_NOT_AVAIL == iDiagonalRef) {
iDiagonalRef = iRefIndex[listIdx][2]; //top-left for 8*8 block(index 1)
index = 2;
}
if (iRef == iDiagonalRef) {
ST32 (iMVP, LD32 (&iMotionVector[listIdx][index][0]));
return;
}
}
PredMv (iMotionVector, iRefIndex, listIdx, iPartIdx, 2, iRef, iMVP);
}
void PredInter16x8Mv (int16_t iMotionVector[LIST_A][30][MV_A], int8_t iRefIndex[LIST_A][30],
int32_t listIdx, int32_t iPartIdx, int8_t iRef, int16_t iMVP[2]) {
if (0 == iPartIdx) {
const int8_t kiTopRef = iRefIndex[listIdx][1];
if (iRef == kiTopRef) {
ST32 (iMVP, LD32 (&iMotionVector[listIdx][1][0]));
return;
}
} else { // 8 == iPartIdx
const int8_t kiLeftRef = iRefIndex[listIdx][18];
if (iRef == kiLeftRef) {
ST32 (iMVP, LD32 (&iMotionVector[listIdx][18][0]));
return;
}
}
PredMv (iMotionVector, iRefIndex, listIdx, iPartIdx, 4, iRef, iMVP);
}
//update iMVs and iRefIndex cache for current MB, only for P_16*16 (SKIP inclusive)
/* can be further optimized */
void UpdateP16x16MotionInfo (PDqLayer pCurDqLayer, int32_t listIdx, int8_t iRef, int16_t iMVs[2]) {
const int16_t kiRef2 = ((uint8_t)iRef << 8) | (uint8_t)iRef;
const int32_t kiMV32 = LD32 (iMVs);
int32_t i;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
for (i = 0; i < 16; i += 4) {
//mb
const uint8_t kuiScan4Idx = g_kuiScan4[i];
const uint8_t kuiScan4IdxPlus4 = 4 + kuiScan4Idx;
if (pCurDqLayer->pDec != NULL) {
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
} else {
ST16 (&pCurDqLayer->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
}
}
}
//update iRefIndex cache for current MB, only for P_16*16 (SKIP inclusive)
/* can be further optimized */
void UpdateP16x16RefIdx (PDqLayer pCurDqLayer, int32_t listIdx, int8_t iRef) {
const int16_t kiRef2 = ((uint8_t)iRef << 8) | (uint8_t)iRef;
int32_t i;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
for (i = 0; i < 16; i += 4) {
//mb
const uint8_t kuiScan4Idx = g_kuiScan4[i];
const uint8_t kuiScan4IdxPlus4 = 4 + kuiScan4Idx;
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
}
}
//update iMVs only cache for current MB, only for P_16*16 (SKIP inclusive)
/* can be further optimized */
void UpdateP16x16MotionOnly (PDqLayer pCurDqLayer, int32_t listIdx, int16_t iMVs[2]) {
const int32_t kiMV32 = LD32 (iMVs);
int32_t i;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
for (i = 0; i < 16; i += 4) {
//mb
const uint8_t kuiScan4Idx = g_kuiScan4[i];
const uint8_t kuiScan4IdxPlus4 = 4 + kuiScan4Idx;
if (pCurDqLayer->pDec != NULL) {
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
} else {
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
}
}
}
//update iRefIndex and iMVs of Mb, only for P16x8
/*need further optimization, mb_cache not work */
void UpdateP16x8MotionInfo (PDqLayer pCurDqLayer, int16_t iMotionVector[LIST_A][30][MV_A],
int8_t iRefIndex[LIST_A][30],
int32_t listIdx, int32_t iPartIdx, int8_t iRef, int16_t iMVs[2]) {
const int16_t kiRef2 = ((uint8_t)iRef << 8) | (uint8_t)iRef;
const int32_t kiMV32 = LD32 (iMVs);
int32_t i;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
for (i = 0; i < 2; i++, iPartIdx += 4) {
const uint8_t kuiScan4Idx = g_kuiScan4[iPartIdx];
const uint8_t kuiScan4IdxPlus4 = 4 + kuiScan4Idx;
const uint8_t kuiCacheIdx = g_kuiCache30ScanIdx[iPartIdx];
const uint8_t kuiCacheIdxPlus6 = 6 + kuiCacheIdx;
//mb
if (pCurDqLayer->pDec != NULL) {
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
} else {
ST16 (&pCurDqLayer->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
}
//cache
ST16 (&iRefIndex[listIdx][kuiCacheIdx ], kiRef2);
ST16 (&iRefIndex[listIdx][kuiCacheIdxPlus6], kiRef2);
ST32 (iMotionVector[listIdx][ kuiCacheIdx ], kiMV32);
ST32 (iMotionVector[listIdx][1 + kuiCacheIdx ], kiMV32);
ST32 (iMotionVector[listIdx][ kuiCacheIdxPlus6], kiMV32);
ST32 (iMotionVector[listIdx][1 + kuiCacheIdxPlus6], kiMV32);
}
}
//update iRefIndex and iMVs of both Mb and Mb_cache, only for P8x16
void UpdateP8x16MotionInfo (PDqLayer pCurDqLayer, int16_t iMotionVector[LIST_A][30][MV_A],
int8_t iRefIndex[LIST_A][30],
int32_t listIdx, int32_t iPartIdx, int8_t iRef, int16_t iMVs[2]) {
const int16_t kiRef2 = ((uint8_t)iRef << 8) | (uint8_t)iRef;
const int32_t kiMV32 = LD32 (iMVs);
int32_t i;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
for (i = 0; i < 2; i++, iPartIdx += 8) {
const uint8_t kuiScan4Idx = g_kuiScan4[iPartIdx];
const uint8_t kuiCacheIdx = g_kuiCache30ScanIdx[iPartIdx];
const uint8_t kuiScan4IdxPlus4 = 4 + kuiScan4Idx;
const uint8_t kuiCacheIdxPlus6 = 6 + kuiCacheIdx;
//mb
if (pCurDqLayer->pDec != NULL) {
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pDec->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
} else {
ST16 (&pCurDqLayer->pRefIndex[listIdx][iMbXy][kuiScan4Idx], kiRef2);
ST16 (&pCurDqLayer->pRefIndex[listIdx][iMbXy][kuiScan4IdxPlus4], kiRef2);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4Idx], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][kuiScan4IdxPlus4], kiMV32);
ST32 (pCurDqLayer->pMv[listIdx][iMbXy][1 + kuiScan4IdxPlus4], kiMV32);
}
//cache
ST16 (&iRefIndex[listIdx][kuiCacheIdx ], kiRef2);
ST16 (&iRefIndex[listIdx][kuiCacheIdxPlus6], kiRef2);
ST32 (iMotionVector[listIdx][ kuiCacheIdx ], kiMV32);
ST32 (iMotionVector[listIdx][1 + kuiCacheIdx ], kiMV32);
ST32 (iMotionVector[listIdx][ kuiCacheIdxPlus6], kiMV32);
ST32 (iMotionVector[listIdx][1 + kuiCacheIdxPlus6], kiMV32);
}
}
void FillSpatialDirect8x8Mv (PDqLayer pCurDqLayer, const int16_t& iIdx8, const int8_t& iPartCount, const int8_t& iPartW,
const SubMbType& subMbType, const bool& bIsLongRef, int16_t pMvDirect[LIST_A][2], int8_t iRef[LIST_A],
int16_t pMotionVector[LIST_A][30][MV_A], int16_t pMvdCache[LIST_A][30][MV_A]) {
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
for (int32_t j = 0; j < iPartCount; j++) {
int8_t iPartIdx = iIdx8 + j * iPartW;
uint8_t iScan4Idx = g_kuiScan4[iPartIdx];
uint8_t iColocIdx = g_kuiScan4[iPartIdx];
uint8_t iCacheIdx = g_kuiCache30ScanIdx[iPartIdx];
int16_t pMV[4] = { 0 };
if (IS_SUB_8x8 (subMbType)) {
* (uint32_t*)pMV = * (uint32_t*)pMvDirect[LIST_0];
ST32 ((pMV + 2), LD32 (pMV));
ST64 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx], LD64 (pMV));
ST64 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx + 4], LD64 (pMV));
ST64 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx + 4], 0);
if (pMotionVector != NULL) {
ST64 (pMotionVector[LIST_0][iCacheIdx], LD64 (pMV));
ST64 (pMotionVector[LIST_0][iCacheIdx + 6], LD64 (pMV));
}
if (pMvdCache != NULL) {
ST64 (pMvdCache[LIST_0][iCacheIdx], 0);
ST64 (pMvdCache[LIST_0][iCacheIdx + 6], 0);
}
* (uint32_t*)pMV = * (uint32_t*)pMvDirect[LIST_1];
ST32 ((pMV + 2), LD32 (pMV));
ST64 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx], LD64 (pMV));
ST64 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx + 4], LD64 (pMV));
ST64 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx + 4], 0);
if (pMotionVector != NULL) {
ST64 (pMotionVector[LIST_1][iCacheIdx], LD64 (pMV));
ST64 (pMotionVector[LIST_1][iCacheIdx + 6], LD64 (pMV));
}
if (pMvdCache != NULL) {
ST64 (pMvdCache[LIST_1][iCacheIdx], 0);
ST64 (pMvdCache[LIST_1][iCacheIdx + 6], 0);
}
} else { //SUB_4x4
* (uint32_t*)pMV = * (uint32_t*)pMvDirect[LIST_0];
ST32 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx], LD32 (pMV));
ST32 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx], 0);
if (pMotionVector != NULL) {
ST32 (pMotionVector[LIST_0][iCacheIdx], LD32 (pMV));
}
if (pMvdCache != NULL) {
ST32 (pMvdCache[LIST_0][iCacheIdx], 0);
}
* (uint32_t*)pMV = * (uint32_t*)pMvDirect[LIST_1];
ST32 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx], LD32 (pMV));
ST32 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx], 0);
if (pMotionVector != NULL) {
ST32 (pMotionVector[LIST_1][iCacheIdx], LD32 (pMV));
}
if (pMvdCache != NULL) {
ST32 (pMvdCache[LIST_1][iCacheIdx], 0);
}
}
if ((* (int32_t*)pMvDirect[LIST_0] | * (int32_t*)pMvDirect[LIST_1])) {
uint32_t uiColZeroFlag = (0 == pCurDqLayer->iColocIntra[iColocIdx]) && !bIsLongRef &&
(pCurDqLayer->iColocRefIndex[LIST_0][iColocIdx] == 0 || (pCurDqLayer->iColocRefIndex[LIST_0][iColocIdx] < 0
&& pCurDqLayer->iColocRefIndex[LIST_1][iColocIdx] == 0));
const int16_t (*mvColoc)[2] = 0 == pCurDqLayer->iColocRefIndex[LIST_0][iColocIdx] ? pCurDqLayer->iColocMv[LIST_0] :
pCurDqLayer->iColocMv[LIST_1];
const int16_t* mv = mvColoc[iColocIdx];
if (IS_SUB_8x8 (subMbType)) {
if (uiColZeroFlag && ((unsigned) (mv[0] + 1) <= 2 && (unsigned) (mv[1] + 1) <= 2)) {
if (iRef[LIST_0] == 0) {
ST64 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx + 4], 0);
ST64 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx + 4], 0);
if (pMotionVector != NULL) {
ST64 (pMotionVector[LIST_0][iCacheIdx], 0);
ST64 (pMotionVector[LIST_0][iCacheIdx + 6], 0);
}
if (pMvdCache != NULL) {
ST64 (pMvdCache[LIST_0][iCacheIdx], 0);
ST64 (pMvdCache[LIST_0][iCacheIdx + 6], 0);
}
}
if (iRef[LIST_1] == 0) {
ST64 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx + 4], 0);
ST64 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx + 4], 0);
if (pMotionVector != NULL) {
ST64 (pMotionVector[LIST_1][iCacheIdx], 0);
ST64 (pMotionVector[LIST_1][iCacheIdx + 6], 0);
}
if (pMvdCache != NULL) {
ST64 (pMvdCache[LIST_1][iCacheIdx], 0);
ST64 (pMvdCache[LIST_1][iCacheIdx + 6], 0);
}
}
}
} else {
if (uiColZeroFlag && ((unsigned) (mv[0] + 1) <= 2 && (unsigned) (mv[1] + 1) <= 2)) {
if (iRef[LIST_0] == 0) {
ST32 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx], 0);
ST32 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx], 0);
if (pMotionVector != NULL) {
ST32 (pMotionVector[LIST_0][iCacheIdx], 0);
}
if (pMvdCache != NULL) {
ST32 (pMvdCache[LIST_0][iCacheIdx], 0);
}
}
if (iRef[LIST_1] == 0) {
ST32 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx], 0);
ST32 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx], 0);
if (pMotionVector != NULL) {
ST32 (pMotionVector[LIST_1][iCacheIdx], 0);
}
if (pMvdCache != NULL) {
ST32 (pMvdCache[LIST_1][iCacheIdx], 0);
}
}
}
}
}
}
}
void FillTemporalDirect8x8Mv (PDqLayer pCurDqLayer, const int16_t& iIdx8, const int8_t& iPartCount,
const int8_t& iPartW,
const SubMbType& subMbType, int8_t iRef[LIST_A], int16_t (*mvColoc)[2], int16_t pMotionVector[LIST_A][30][MV_A],
int16_t pMvdCache[LIST_A][30][MV_A]) {
PSlice pSlice = &pCurDqLayer->sLayerInfo.sSliceInLayer;
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
int16_t pMvDirect[LIST_A][2] = { { 0, 0 }, { 0, 0 } };
for (int32_t j = 0; j < iPartCount; j++) {
int8_t iPartIdx = iIdx8 + j * iPartW;
uint8_t iScan4Idx = g_kuiScan4[iPartIdx];
uint8_t iColocIdx = g_kuiScan4[iPartIdx];
uint8_t iCacheIdx = g_kuiCache30ScanIdx[iPartIdx];
int16_t* mv = mvColoc[iColocIdx];
int16_t pMV[4] = { 0 };
if (IS_SUB_8x8 (subMbType)) {
if (!pCurDqLayer->iColocIntra[iColocIdx]) {
pMvDirect[LIST_0][0] = (pSlice->iMvScale[LIST_0][iRef[LIST_0]] * mv[0] + 128) >> 8;
pMvDirect[LIST_0][1] = (pSlice->iMvScale[LIST_0][iRef[LIST_0]] * mv[1] + 128) >> 8;
}
ST32 (pMV, LD32 (pMvDirect[LIST_0]));
ST32 ((pMV + 2), LD32 (pMvDirect[LIST_0]));
ST64 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx], LD64 (pMV));
ST64 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx + 4], LD64 (pMV));
ST64 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx + 4], 0);
if (pMotionVector != NULL) {
ST64 (pMotionVector[LIST_0][iCacheIdx], LD64 (pMV));
ST64 (pMotionVector[LIST_0][iCacheIdx + 6], LD64 (pMV));
}
if (pMvdCache != NULL) {
ST64 (pMvdCache[LIST_0][iCacheIdx], 0);
ST64 (pMvdCache[LIST_0][iCacheIdx + 6], 0);
}
if (!pCurDqLayer->iColocIntra[g_kuiScan4[iIdx8]]) {
pMvDirect[LIST_1][0] = pMvDirect[LIST_0][0] - mv[0];
pMvDirect[LIST_1][1] = pMvDirect[LIST_0][1] - mv[1];
}
ST32 (pMV, LD32 (pMvDirect[LIST_1]));
ST32 ((pMV + 2), LD32 (pMvDirect[LIST_1]));
ST64 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx], LD64 (pMV));
ST64 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx + 4], LD64 (pMV));
ST64 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx], 0);
ST64 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx + 4], 0);
if (pMotionVector != NULL) {
ST64 (pMotionVector[LIST_1][iCacheIdx], LD64 (pMV));
ST64 (pMotionVector[LIST_1][iCacheIdx + 6], LD64 (pMV));
}
if (pMvdCache != NULL) {
ST64 (pMvdCache[LIST_1][iCacheIdx], 0);
ST64 (pMvdCache[LIST_1][iCacheIdx + 6], 0);
}
} else { //SUB_4x4
if (!pCurDqLayer->iColocIntra[iColocIdx]) {
pMvDirect[LIST_0][0] = (pSlice->iMvScale[LIST_0][iRef[LIST_0]] * mv[0] + 128) >> 8;
pMvDirect[LIST_0][1] = (pSlice->iMvScale[LIST_0][iRef[LIST_0]] * mv[1] + 128) >> 8;
}
ST32 (pCurDqLayer->pDec->pMv[LIST_0][iMbXy][iScan4Idx], LD32 (pMvDirect[LIST_0]));
ST32 (pCurDqLayer->pMvd[LIST_0][iMbXy][iScan4Idx], 0);
if (pMotionVector != NULL) {
ST32 (pMotionVector[LIST_0][iCacheIdx], LD32 (pMvDirect[LIST_0]));
}
if (pMvdCache != NULL) {
ST32 (pMvdCache[LIST_0][iCacheIdx], 0);
}
if (!pCurDqLayer->iColocIntra[iColocIdx]) {
pMvDirect[LIST_1][0] = pMvDirect[LIST_0][0] - mv[0];
pMvDirect[LIST_1][1] = pMvDirect[LIST_0][1] - mv[1];
}
ST32 (pCurDqLayer->pDec->pMv[LIST_1][iMbXy][iScan4Idx], LD32 (pMvDirect[LIST_1]));
ST32 (pCurDqLayer->pMvd[LIST_1][iMbXy][iScan4Idx], 0);
if (pMotionVector != NULL) {
ST32 (pMotionVector[LIST_1][iCacheIdx], LD32 (pMvDirect[LIST_1]));
}
if (pMvdCache != NULL) {
ST32 (pMvdCache[LIST_1][iCacheIdx], 0);
}
}
}
}
int8_t MapColToList0 (PWelsDecoderContext& pCtx, const int8_t& colocRefIndexL0,
const int32_t& ref0Count) { //ISO/IEC 14496-10:2009(E) (8-193)
//When reference is lost, this function must be skipped.
if ((pCtx->iErrorCode & dsRefLost) == dsRefLost) {
return 0;
}
PPicture pic1 = pCtx->sRefPic.pRefList[LIST_1][0];
if (pic1 && pic1->pRefPic[LIST_0][colocRefIndexL0]) {
const int32_t iFramePoc = pic1->pRefPic[LIST_0][colocRefIndexL0]->iFramePoc;
for (int32_t i = 0; i < ref0Count; i++) {
if (pCtx->sRefPic.pRefList[LIST_0][i]->iFramePoc == iFramePoc) {
return i;
}
}
}
return 0;
}
void Update8x8RefIdx (PDqLayer& pCurDqLayer, const int16_t& iPartIdx, const int32_t& listIdx, const int8_t& iRef) {
int32_t iMbXy = pCurDqLayer->iMbXyIndex;
const uint8_t iScan4Idx = g_kuiScan4[iPartIdx];
pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][iScan4Idx] = pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][iScan4Idx + 1] =
pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][iScan4Idx + 4] = pCurDqLayer->pDec->pRefIndex[listIdx][iMbXy][iScan4Idx +
5] = iRef;
}
} // namespace WelsDec