/* ------------------------------------------------------------------ * Copyright (C) 1998-2009 PacketVideo * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either * express or implied. * See the License for the specific language governing permissions * and limitations under the License. * ------------------------------------------------------------------- */ #include #include "avclib_common.h" #define MAX_QP 51 #define MB_BLOCK_SIZE 16 // NOTE: these 3 tables are for funtion GetStrength() only const static int ININT_STRENGTH[4] = {0x04040404, 0x03030303, 0x03030303, 0x03030303}; // NOTE: these 3 tables are for funtion EdgeLoop() only // NOTE: to change the tables below for instance when the QP doubling is changed from 6 to 8 values const static int ALPHA_TABLE[52] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4, 4, 5, 6, 7, 8, 9, 10, 12, 13, 15, 17, 20, 22, 25, 28, 32, 36, 40, 45, 50, 56, 63, 71, 80, 90, 101, 113, 127, 144, 162, 182, 203, 226, 255, 255} ; const static int BETA_TABLE[52] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14, 15, 15, 16, 16, 17, 17, 18, 18} ; const static int CLIP_TAB[52][5] = { { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 0, 0}, { 0, 0, 0, 1, 1}, { 0, 0, 0, 1, 1}, { 0, 0, 0, 1, 1}, { 0, 0, 0, 1, 1}, { 0, 0, 1, 1, 1}, { 0, 0, 1, 1, 1}, { 0, 1, 1, 1, 1}, { 0, 1, 1, 1, 1}, { 0, 1, 1, 1, 1}, { 0, 1, 1, 1, 1}, { 0, 1, 1, 2, 2}, { 0, 1, 1, 2, 2}, { 0, 1, 1, 2, 2}, { 0, 1, 1, 2, 2}, { 0, 1, 2, 3, 3}, { 0, 1, 2, 3, 3}, { 0, 2, 2, 3, 3}, { 0, 2, 2, 4, 4}, { 0, 2, 3, 4, 4}, { 0, 2, 3, 4, 4}, { 0, 3, 3, 5, 5}, { 0, 3, 4, 6, 6}, { 0, 3, 4, 6, 6}, { 0, 4, 5, 7, 7}, { 0, 4, 5, 8, 8}, { 0, 4, 6, 9, 9}, { 0, 5, 7, 10, 10}, { 0, 6, 8, 11, 11}, { 0, 6, 8, 13, 13}, { 0, 7, 10, 14, 14}, { 0, 8, 11, 16, 16}, { 0, 9, 12, 18, 18}, { 0, 10, 13, 20, 20}, { 0, 11, 15, 23, 23}, { 0, 13, 17, 25, 25} }; // NOTE: this table is only QP clipping, index = QP + video->FilterOffsetA/B, clipped to [0, 51] // video->FilterOffsetA/B is in {-12, 12] const static int QP_CLIP_TAB[76] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // [-12, 0] 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, // [1, 51] 51, 51, 51, 51, 51, 51, 51, 51, 51, 51, 51, 51 // [52,63] }; static void DeblockMb(AVCCommonObj *video, int mb_x, int mb_y, uint8 *SrcY, uint8 *SrcU, uint8 *SrcV); //static void GetStrength(AVCCommonObj *video, uint8 *Strength, AVCMacroblock* MbP, AVCMacroblock* MbQ, int dir, int edge); static void GetStrength_Edge0(uint8 *Strength, AVCMacroblock* MbP, AVCMacroblock* MbQ, int dir); static void GetStrength_VerticalEdges(uint8 *Strength, AVCMacroblock* MbQ); static void GetStrength_HorizontalEdges(uint8 Strength[12], AVCMacroblock* MbQ); static void EdgeLoop_Luma_vertical(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch); static void EdgeLoop_Luma_horizontal(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch); static void EdgeLoop_Chroma_vertical(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch); static void EdgeLoop_Chroma_horizontal(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch); /* ***************************************************************************************** * \brief Filter all macroblocks in order of increasing macroblock address. ***************************************************************************************** */ OSCL_EXPORT_REF AVCStatus DeblockPicture(AVCCommonObj *video) { uint i, j; int pitch = video->currPic->pitch, pitch_c, width; uint8 *SrcY, *SrcU, *SrcV; SrcY = video->currPic->Sl; // pointers to source SrcU = video->currPic->Scb; SrcV = video->currPic->Scr; pitch_c = pitch >> 1; width = video->currPic->width; for (i = 0; i < video->PicHeightInMbs; i++) { for (j = 0; j < video->PicWidthInMbs; j++) { DeblockMb(video, j, i, SrcY, SrcU, SrcV); // update SrcY, SrcU, SrcV SrcY += MB_BLOCK_SIZE; SrcU += (MB_BLOCK_SIZE >> 1); SrcV += (MB_BLOCK_SIZE >> 1); } SrcY += ((pitch << 4) - width); SrcU += ((pitch_c << 3) - (width >> 1)); SrcV += ((pitch_c << 3) - (width >> 1)); } return AVC_SUCCESS; } #ifdef MB_BASED_DEBLOCK /* ***************************************************************************************** * \brief Filter one macroblocks in a fast macroblock memory and copy it to frame ***************************************************************************************** */ void MBInLoopDeblock(AVCCommonObj *video) { AVCPictureData *currPic = video->currPic; #ifdef USE_PRED_BLOCK uint8 *predCb, *predCr, *pred_block; int i, j, dst_width, dst_height, dst_widthc, dst_heightc; #endif int pitch = currPic->pitch; int x_pos = video->mb_x; int y_pos = video->mb_y; uint8 *curL, *curCb, *curCr; int offset; offset = (y_pos << 4) * pitch; curL = currPic->Sl + offset + (x_pos << 4); offset >>= 2; offset += (x_pos << 3); curCb = currPic->Scb + offset; curCr = currPic->Scr + offset; #ifdef USE_PRED_BLOCK pred_block = video->pred; /* 1. copy neighboring pixels from frame to the video->pred_block */ if (y_pos) /* not the 0th row */ { /* copy to the top 4 lines of the macroblock */ curL -= (pitch << 2); /* go back 4 lines */ memcpy(pred_block + 4, curL, 16); curL += pitch; memcpy(pred_block + 24, curL, 16); curL += pitch; memcpy(pred_block + 44, curL, 16); curL += pitch; memcpy(pred_block + 64, curL, 16); curL += pitch; curCb -= (pitch << 1); /* go back 4 lines chroma */ curCr -= (pitch << 1); pred_block += 400; memcpy(pred_block + 4, curCb, 8); curCb += (pitch >> 1); memcpy(pred_block + 16, curCb, 8); curCb += (pitch >> 1); memcpy(pred_block + 28, curCb, 8); curCb += (pitch >> 1); memcpy(pred_block + 40, curCb, 8); curCb += (pitch >> 1); pred_block += 144; memcpy(pred_block + 4, curCr, 8); curCr += (pitch >> 1); memcpy(pred_block + 16, curCr, 8); curCr += (pitch >> 1); memcpy(pred_block + 28, curCr, 8); curCr += (pitch >> 1); memcpy(pred_block + 40, curCr, 8); curCr += (pitch >> 1); pred_block = video->pred; } /* 2. perform deblocking. */ DeblockMb(video, x_pos, y_pos, pred_block + 84, pred_block + 452, pred_block + 596); /* 3. copy it back to the frame and update pred_block */ predCb = pred_block + 400; predCr = predCb + 144; /* find the range of the block inside pred_block to be copied back */ if (y_pos) /* the first row */ { curL -= (pitch << 2); curCb -= (pitch << 1); curCr -= (pitch << 1); dst_height = 20; dst_heightc = 12; } else { pred_block += 80; predCb += 48; predCr += 48; dst_height = 16; dst_heightc = 8; } if (x_pos) /* find the width */ { curL -= 4; curCb -= 4; curCr -= 4; if (x_pos == (int)(video->PicWidthInMbs - 1)) { dst_width = 20; dst_widthc = 12; } else { dst_width = 16; dst_widthc = 8; } } else { pred_block += 4; predCb += 4; predCr += 4; dst_width = 12; dst_widthc = 4; } /* perform copy */ for (j = 0; j < dst_height; j++) { memcpy(curL, pred_block, dst_width); curL += pitch; pred_block += 20; } for (j = 0; j < dst_heightc; j++) { memcpy(curCb, predCb, dst_widthc); memcpy(curCr, predCr, dst_widthc); curCb += (pitch >> 1); curCr += (pitch >> 1); predCb += 12; predCr += 12; } if (x_pos != (int)(video->PicWidthInMbs - 1)) /* now copy from the right-most 4 columns to the left-most 4 columns */ { pred_block = video->pred; for (i = 0; i < 20; i += 4) { *((uint32*)pred_block) = *((uint32*)(pred_block + 16)); pred_block += 20; *((uint32*)pred_block) = *((uint32*)(pred_block + 16)); pred_block += 20; *((uint32*)pred_block) = *((uint32*)(pred_block + 16)); pred_block += 20; *((uint32*)pred_block) = *((uint32*)(pred_block + 16)); pred_block += 20; } for (i = 0; i < 24; i += 4) { *((uint32*)pred_block) = *((uint32*)(pred_block + 8)); pred_block += 12; *((uint32*)pred_block) = *((uint32*)(pred_block + 8)); pred_block += 12; *((uint32*)pred_block) = *((uint32*)(pred_block + 8)); pred_block += 12; *((uint32*)pred_block) = *((uint32*)(pred_block + 8)); pred_block += 12; } } #else DeblockMb(video, x_pos, y_pos, curL, curCb, curCr); #endif return ; } #endif /* ***************************************************************************************** * \brief Deblocking filter for one macroblock. ***************************************************************************************** */ void DeblockMb(AVCCommonObj *video, int mb_x, int mb_y, uint8 *SrcY, uint8 *SrcU, uint8 *SrcV) { AVCMacroblock *MbP, *MbQ; int edge, QP, QPC; int filterLeftMbEdgeFlag = (mb_x != 0); int filterTopMbEdgeFlag = (mb_y != 0); int pitch = video->currPic->pitch; int indexA, indexB; int *tmp; int Alpha, Beta, Alpha_c, Beta_c; int mbNum = mb_y * video->PicWidthInMbs + mb_x; int *clipTable, *clipTable_c, *qp_clip_tab; uint8 Strength[16]; void* str; MbQ = &(video->mblock[mbNum]); // current Mb // If filter is disabled, return if (video->sliceHdr->disable_deblocking_filter_idc == 1) return; if (video->sliceHdr->disable_deblocking_filter_idc == 2) { // don't filter at slice boundaries filterLeftMbEdgeFlag = mb_is_available(video->mblock, video->PicSizeInMbs, mbNum - 1, mbNum); filterTopMbEdgeFlag = mb_is_available(video->mblock, video->PicSizeInMbs, mbNum - video->PicWidthInMbs, mbNum); } /* NOTE: edge=0 and edge=1~3 are separate cases because of the difference of MbP, index A and indexB calculation */ /* for edge = 1~3, MbP, indexA and indexB remain the same, and thus there is no need to re-calculate them for each edge */ qp_clip_tab = (int *)QP_CLIP_TAB + 12; /* 1.VERTICAL EDGE + MB BOUNDARY (edge = 0) */ if (filterLeftMbEdgeFlag) { MbP = MbQ - 1; //GetStrength(video, Strength, MbP, MbQ, 0, 0); // Strength for 4 blks in 1 stripe, 0 => vertical edge GetStrength_Edge0(Strength, MbP, MbQ, 0); str = (void*)Strength; //de-ref type-punned pointer fix if (*((uint32*)str)) // only if one of the 4 Strength bytes is != 0 { QP = (MbP->QPy + MbQ->QPy + 1) >> 1; // Average QP of the two blocks; indexA = QP + video->FilterOffsetA; indexB = QP + video->FilterOffsetB; indexA = qp_clip_tab[indexA]; // IClip(0, MAX_QP, QP+video->FilterOffsetA) indexB = qp_clip_tab[indexB]; // IClip(0, MAX_QP, QP+video->FilterOffsetB) Alpha = ALPHA_TABLE[indexA]; Beta = BETA_TABLE[indexB]; clipTable = (int *) CLIP_TAB[indexA]; if (Alpha > 0 && Beta > 0) #ifdef USE_PRED_BLOCK EdgeLoop_Luma_vertical(SrcY, Strength, Alpha, Beta, clipTable, 20); #else EdgeLoop_Luma_vertical(SrcY, Strength, Alpha, Beta, clipTable, pitch); #endif QPC = (MbP->QPc + MbQ->QPc + 1) >> 1; indexA = QPC + video->FilterOffsetA; indexB = QPC + video->FilterOffsetB; indexA = qp_clip_tab[indexA]; // IClip(0, MAX_QP, QP+video->FilterOffsetA) indexB = qp_clip_tab[indexB]; // IClip(0, MAX_QP, QP+video->FilterOffsetB) Alpha = ALPHA_TABLE[indexA]; Beta = BETA_TABLE[indexB]; clipTable = (int *) CLIP_TAB[indexA]; if (Alpha > 0 && Beta > 0) { #ifdef USE_PRED_BLOCK EdgeLoop_Chroma_vertical(SrcU, Strength, Alpha, Beta, clipTable, 12); EdgeLoop_Chroma_vertical(SrcV, Strength, Alpha, Beta, clipTable, 12); #else EdgeLoop_Chroma_vertical(SrcU, Strength, Alpha, Beta, clipTable, pitch >> 1); EdgeLoop_Chroma_vertical(SrcV, Strength, Alpha, Beta, clipTable, pitch >> 1); #endif } } } /* end of: if(filterLeftMbEdgeFlag) */ /* 2.VERTICAL EDGE (no boundary), the edges are all inside a MB */ /* First calculate the necesary parameters all at once, outside the loop */ MbP = MbQ; indexA = MbQ->QPy + video->FilterOffsetA; indexB = MbQ->QPy + video->FilterOffsetB; // index indexA = qp_clip_tab[indexA]; // IClip(0, MAX_QP, QP+video->FilterOffsetA) indexB = qp_clip_tab[indexB]; // IClip(0, MAX_QP, QP+video->FilterOffsetB) Alpha = ALPHA_TABLE[indexA]; Beta = BETA_TABLE[indexB]; clipTable = (int *)CLIP_TAB[indexA]; /* Save Alpha, Beta and clipTable for future use, with the obselete variables filterLeftMbEdgeFlag, mbNum amd tmp */ filterLeftMbEdgeFlag = Alpha; mbNum = Beta; tmp = clipTable; indexA = MbQ->QPc + video->FilterOffsetA; indexB = MbQ->QPc + video->FilterOffsetB; indexA = qp_clip_tab[indexA]; // IClip(0, MAX_QP, QP+video->FilterOffsetA) indexB = qp_clip_tab[indexB]; // IClip(0, MAX_QP, QP+video->FilterOffsetB) Alpha_c = ALPHA_TABLE[indexA]; Beta_c = BETA_TABLE[indexB]; clipTable_c = (int *)CLIP_TAB[indexA]; GetStrength_VerticalEdges(Strength + 4, MbQ); // Strength for 4 blks in 1 stripe, 0 => vertical edge for (edge = 1; edge < 4; edge++) // 4 vertical strips of 16 pel { //GetStrength_VerticalEdges(video, Strength, MbP, MbQ, 0, edge); // Strength for 4 blks in 1 stripe, 0 => vertical edge if (*((int*)(Strength + (edge << 2)))) // only if one of the 4 Strength bytes is != 0 { if (Alpha > 0 && Beta > 0) #ifdef USE_PRED_BLOCK EdgeLoop_Luma_vertical(SrcY + (edge << 2), Strength + (edge << 2), Alpha, Beta, clipTable, 20); #else EdgeLoop_Luma_vertical(SrcY + (edge << 2), Strength + (edge << 2), Alpha, Beta, clipTable, pitch); #endif if (!(edge & 1) && Alpha_c > 0 && Beta_c > 0) { #ifdef USE_PRED_BLOCK EdgeLoop_Chroma_vertical(SrcU + (edge << 1), Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, 12); EdgeLoop_Chroma_vertical(SrcV + (edge << 1), Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, 12); #else EdgeLoop_Chroma_vertical(SrcU + (edge << 1), Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, pitch >> 1); EdgeLoop_Chroma_vertical(SrcV + (edge << 1), Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, pitch >> 1); #endif } } } //end edge /* 3.HORIZONTAL EDGE + MB BOUNDARY (edge = 0) */ if (filterTopMbEdgeFlag) { MbP = MbQ - video->PicWidthInMbs; //GetStrength(video, Strength, MbP, MbQ, 1, 0); // Strength for 4 blks in 1 stripe, 0 => vertical edge GetStrength_Edge0(Strength, MbP, MbQ, 1); str = (void*)Strength; //de-ref type-punned pointer fix if (*((uint32*)str)) // only if one of the 4 Strength bytes is != 0 { QP = (MbP->QPy + MbQ->QPy + 1) >> 1; // Average QP of the two blocks; indexA = QP + video->FilterOffsetA; indexB = QP + video->FilterOffsetB; indexA = qp_clip_tab[indexA]; // IClip(0, MAX_QP, QP+video->FilterOffsetA) indexB = qp_clip_tab[indexB]; // IClip(0, MAX_QP, QP+video->FilterOffsetB) Alpha = ALPHA_TABLE[indexA]; Beta = BETA_TABLE[indexB]; clipTable = (int *)CLIP_TAB[indexA]; if (Alpha > 0 && Beta > 0) { #ifdef USE_PRED_BLOCK EdgeLoop_Luma_horizontal(SrcY, Strength, Alpha, Beta, clipTable, 20); #else EdgeLoop_Luma_horizontal(SrcY, Strength, Alpha, Beta, clipTable, pitch); #endif } QPC = (MbP->QPc + MbQ->QPc + 1) >> 1; indexA = QPC + video->FilterOffsetA; indexB = QPC + video->FilterOffsetB; indexA = qp_clip_tab[indexA]; // IClip(0, MAX_QP, QP+video->FilterOffsetA) indexB = qp_clip_tab[indexB]; // IClip(0, MAX_QP, QP+video->FilterOffsetB) Alpha = ALPHA_TABLE[indexA]; Beta = BETA_TABLE[indexB]; clipTable = (int *)CLIP_TAB[indexA]; if (Alpha > 0 && Beta > 0) { #ifdef USE_PRED_BLOCK EdgeLoop_Chroma_horizontal(SrcU, Strength, Alpha, Beta, clipTable, 12); EdgeLoop_Chroma_horizontal(SrcV, Strength, Alpha, Beta, clipTable, 12); #else EdgeLoop_Chroma_horizontal(SrcU, Strength, Alpha, Beta, clipTable, pitch >> 1); EdgeLoop_Chroma_horizontal(SrcV, Strength, Alpha, Beta, clipTable, pitch >> 1); #endif } } } /* end of: if(filterTopMbEdgeFlag) */ /* 4.HORIZONTAL EDGE (no boundary), the edges are inside a MB */ MbP = MbQ; /* Recover Alpha, Beta and clipTable for edge!=0 with the variables filterLeftMbEdgeFlag, mbNum and tmp */ /* Note that Alpha_c, Beta_c and clipTable_c for chroma is already calculated */ Alpha = filterLeftMbEdgeFlag; Beta = mbNum; clipTable = tmp; GetStrength_HorizontalEdges(Strength + 4, MbQ); // Strength for 4 blks in 1 stripe, 0 => vertical edge for (edge = 1; edge < 4; edge++) // 4 horicontal strips of 16 pel { //GetStrength(video, Strength, MbP, MbQ, 1, edge); // Strength for 4 blks in 1 stripe 1 => horizontal edge if (*((int*)(Strength + (edge << 2)))) // only if one of the 4 Strength bytes is != 0 { if (Alpha > 0 && Beta > 0) { #ifdef USE_PRED_BLOCK EdgeLoop_Luma_horizontal(SrcY + (edge << 2)*20, Strength + (edge << 2), Alpha, Beta, clipTable, 20); #else EdgeLoop_Luma_horizontal(SrcY + (edge << 2)*pitch, Strength + (edge << 2), Alpha, Beta, clipTable, pitch); #endif } if (!(edge & 1) && Alpha_c > 0 && Beta_c > 0) { #ifdef USE_PRED_BLOCK EdgeLoop_Chroma_horizontal(SrcU + (edge << 1)*12, Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, 12); EdgeLoop_Chroma_horizontal(SrcV + (edge << 1)*12, Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, 12); #else EdgeLoop_Chroma_horizontal(SrcU + (edge << 1)*(pitch >> 1), Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, pitch >> 1); EdgeLoop_Chroma_horizontal(SrcV + (edge << 1)*(pitch >> 1), Strength + (edge << 2), Alpha_c, Beta_c, clipTable_c, pitch >> 1); #endif } } } //end edge return; } /* ***************************************************************************************************** * \brief returns a buffer of 4 Strength values for one stripe in a mb (for different Frame types) ***************************************************************************************************** */ void GetStrength_Edge0(uint8 *Strength, AVCMacroblock* MbP, AVCMacroblock* MbQ, int dir) { int tmp; int16 *ptrQ, *ptrP; void* vptr; uint8 *pStrength; void* refIdx; if (MbP->mbMode == AVC_I4 || MbP->mbMode == AVC_I16 || MbQ->mbMode == AVC_I4 || MbQ->mbMode == AVC_I16) { *((int*)Strength) = ININT_STRENGTH[0]; // Start with Strength=3. or Strength=4 for Mb-edge } else // if not intra or SP-frame { *((int*)Strength) = 0; if (dir == 0) // Vertical Edge 0 { //1. Check the ref_frame_id refIdx = (void*) MbQ->RefIdx; //de-ref type-punned pointer fix ptrQ = (int16*)refIdx; refIdx = (void*)MbP->RefIdx; //de-ref type-punned pointer fix ptrP = (int16*)refIdx; pStrength = Strength; if (ptrQ[0] != ptrP[1]) pStrength[0] = 1; if (ptrQ[2] != ptrP[3]) pStrength[2] = 1; pStrength[1] = pStrength[0]; pStrength[3] = pStrength[2]; //2. Check the non-zero coeff blocks (4x4) if (MbQ->nz_coeff[0] != 0 || MbP->nz_coeff[3] != 0) pStrength[0] = 2; if (MbQ->nz_coeff[4] != 0 || MbP->nz_coeff[7] != 0) pStrength[1] = 2; if (MbQ->nz_coeff[8] != 0 || MbP->nz_coeff[11] != 0) pStrength[2] = 2; if (MbQ->nz_coeff[12] != 0 || MbP->nz_coeff[15] != 0) pStrength[3] = 2; //3. Only need to check the mv difference vptr = (void*)MbQ->mvL0; // for deref type-punned pointer ptrQ = (int16*)vptr; ptrP = (int16*)(MbP->mvL0 + 3); // points to 4x4 block #3 (the 4th column) // 1st blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pStrength++; ptrQ += 8; ptrP += 8; // 2nd blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pStrength++; ptrQ += 8; ptrP += 8; // 3rd blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pStrength++; ptrQ += 8; ptrP += 8; // 4th blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } } else // Horizontal Edge 0 { //1. Check the ref_frame_id refIdx = (void*)MbQ->RefIdx; //de-ref type-punned pointer ptrQ = (int16*)refIdx; refIdx = (void*)MbP->RefIdx; //de-ref type-punned pointer ptrP = (int16*)refIdx; pStrength = Strength; if (ptrQ[0] != ptrP[2]) pStrength[0] = 1; if (ptrQ[1] != ptrP[3]) pStrength[2] = 1; pStrength[1] = pStrength[0]; pStrength[3] = pStrength[2]; //2. Check the non-zero coeff blocks (4x4) if (MbQ->nz_coeff[0] != 0 || MbP->nz_coeff[12] != 0) pStrength[0] = 2; if (MbQ->nz_coeff[1] != 0 || MbP->nz_coeff[13] != 0) pStrength[1] = 2; if (MbQ->nz_coeff[2] != 0 || MbP->nz_coeff[14] != 0) pStrength[2] = 2; if (MbQ->nz_coeff[3] != 0 || MbP->nz_coeff[15] != 0) pStrength[3] = 2; //3. Only need to check the mv difference vptr = (void*)MbQ->mvL0; ptrQ = (int16*)vptr; ptrP = (int16*)(MbP->mvL0 + 12); // points to 4x4 block #12 (the 4th row) // 1st blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pStrength++; ptrQ += 2; ptrP += 2; // 2nd blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pStrength++; ptrQ += 2; ptrP += 2; // 3rd blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pStrength++; ptrQ += 2; ptrP += 2; // 4th blk if (*pStrength == 0) { // check |mv difference| >= 4 tmp = *ptrQ++ - *ptrP++; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *ptrQ-- - *ptrP--; if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } } /* end of: else if(dir == 0) */ } /* end of: if( !(MbP->mbMode == AVC_I4 ...) */ } void GetStrength_VerticalEdges(uint8 *Strength, AVCMacroblock* MbQ) { int idx, tmp; int16 *ptr, *pmvx, *pmvy; uint8 *pnz; uint8 *pStrength, *pStr; void* refIdx; if (MbQ->mbMode == AVC_I4 || MbQ->mbMode == AVC_I16) { *((int*)Strength) = ININT_STRENGTH[1]; // Start with Strength=3. or Strength=4 for Mb-edge *((int*)(Strength + 4)) = ININT_STRENGTH[2]; *((int*)(Strength + 8)) = ININT_STRENGTH[3]; } else // Not intra or SP-frame { *((int*)Strength) = 0; // for non-intra MB, strength = 0, 1 or 2. *((int*)(Strength + 4)) = 0; *((int*)(Strength + 8)) = 0; //1. Check the ref_frame_id refIdx = (void*)MbQ->RefIdx; //de-ref type-punned pointer fix ptr = (int16*)refIdx; pStrength = Strength; if (ptr[0] != ptr[1]) pStrength[4] = 1; if (ptr[2] != ptr[3]) pStrength[6] = 1; pStrength[5] = pStrength[4]; pStrength[7] = pStrength[6]; //2. Check the nz_coeff block and mv difference pmvx = (int16*)(MbQ->mvL0 + 1); // points to 4x4 block #1,not #0 pmvy = pmvx + 1; for (idx = 0; idx < 4; idx += 2) // unroll the loop, make 4 iterations to 2 { // first/third row : 1,2,3 or 9,10,12 // Strength = 2 for a whole row pnz = MbQ->nz_coeff + (idx << 2); if (*pnz++ != 0) *pStrength = 2; if (*pnz++ != 0) { *pStrength = 2; *(pStrength + 4) = 2; } if (*pnz++ != 0) { *(pStrength + 4) = 2; *(pStrength + 8) = 2; } if (*pnz != 0) *(pStrength + 8) = 2; // Then Strength = 1 if (*pStrength == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *pmvy - *(pmvy - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pmvx += 2; pmvy += 2; pStr = pStrength + 4; if (*pStr == 0) { //check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } pmvx += 2; pmvy += 2; pStr = pStrength + 8; if (*pStr == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } // Second/fourth row: 5,6,7 or 14,15,16 // Strength = 2 for a whole row pnz = MbQ->nz_coeff + ((idx + 1) << 2); if (*pnz++ != 0) *(pStrength + 1) = 2; if (*pnz++ != 0) { *(pStrength + 1) = 2; *(pStrength + 5) = 2; } if (*pnz++ != 0) { *(pStrength + 5) = 2; *(pStrength + 9) = 2; } if (*pnz != 0) *(pStrength + 9) = 2; // Then Strength = 1 pmvx += 4; pmvy += 4; pStr = pStrength + 1; if (*pStr == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } pmvx += 2; pmvy += 2; pStr = pStrength + 5; if (*pStr == 0) { //check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } pmvx += 2; pmvy += 2; pStr = pStrength + 9; if (*pStr == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 2); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } // update some variables for the next two rows pmvx += 4; pmvy += 4; pStrength += 2; } /* end of: for(idx=0; idx<2; idx++) */ } /* end of: else if( MbQ->mbMode == AVC_I4 ...) */ } void GetStrength_HorizontalEdges(uint8 Strength[12], AVCMacroblock* MbQ) { int idx, tmp; int16 *ptr, *pmvx, *pmvy; uint8 *pStrength, *pStr; void* refIdx; if (MbQ->mbMode == AVC_I4 || MbQ->mbMode == AVC_I16) { *((int*)Strength) = ININT_STRENGTH[1]; // Start with Strength=3. or Strength=4 for Mb-edge *((int*)(Strength + 4)) = ININT_STRENGTH[2]; *((int*)(Strength + 8)) = ININT_STRENGTH[3]; } else // Not intra or SP-frame { *((int*)Strength) = 0; // for non-intra MB, strength = 0, 1 or 2. *((int*)(Strength + 4)) = 0; // for non-intra MB, strength = 0, 1 or 2. *((int*)(Strength + 8)) = 0; // for non-intra MB, strength = 0, 1 or 2. //1. Check the ref_frame_id refIdx = (void*) MbQ->RefIdx; // de-ref type-punned fix ptr = (int16*) refIdx; pStrength = Strength; if (ptr[0] != ptr[2]) pStrength[4] = 1; if (ptr[1] != ptr[3]) pStrength[6] = 1; pStrength[5] = pStrength[4]; pStrength[7] = pStrength[6]; //2. Check the nz_coeff block and mv difference pmvx = (int16*)(MbQ->mvL0 + 4); // points to 4x4 block #4,not #0 pmvy = pmvx + 1; for (idx = 0; idx < 4; idx += 2) // unroll the loop, make 4 iterations to 2 { // first/third row : 1,2,3 or 9,10,12 // Strength = 2 for a whole row if (MbQ->nz_coeff[idx] != 0) *pStrength = 2; if (MbQ->nz_coeff[4+idx] != 0) { *pStrength = 2; *(pStrength + 4) = 2; } if (MbQ->nz_coeff[8+idx] != 0) { *(pStrength + 4) = 2; *(pStrength + 8) = 2; } if (MbQ->nz_coeff[12+idx] != 0) *(pStrength + 8) = 2; // Then Strength = 1 if (*pStrength == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; tmp = *pmvy - *(pmvy - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStrength = 1; } pmvx += 8; pmvy += 8; pStr = pStrength + 4; if (*pStr == 0) { //check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } pmvx += 8; pmvy += 8; pStr = pStrength + 8; if (*pStr == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } // Second/fourth row: 5,6,7 or 14,15,16 // Strength = 2 for a whole row if (MbQ->nz_coeff[idx+1] != 0) *(pStrength + 1) = 2; if (MbQ->nz_coeff[4+idx+1] != 0) { *(pStrength + 1) = 2; *(pStrength + 5) = 2; } if (MbQ->nz_coeff[8+idx+1] != 0) { *(pStrength + 5) = 2; *(pStrength + 9) = 2; } if (MbQ->nz_coeff[12+idx+1] != 0) *(pStrength + 9) = 2; // Then Strength = 1 pmvx -= 14; pmvy -= 14; // -14 = -16 + 2 pStr = pStrength + 1; if (*pStr == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } pmvx += 8; pmvy += 8; pStr = pStrength + 5; if (*pStr == 0) { //check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } pmvx += 8; pmvy += 8; pStr = pStrength + 9; if (*pStr == 0) { //within the same 8x8 block, no need to check the reference id //only need to check the |mv difference| >= 4 tmp = *pmvx - *(pmvx - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; tmp = *pmvy - *(pmvy - 8); if (tmp < 0) tmp = -tmp; if (tmp >= 4) *pStr = 1; } // update some variables for the next two rows pmvx -= 14; pmvy -= 14; // -14 = -16 + 2 pStrength += 2; } /* end of: for(idx=0; idx<2; idx++) */ } /* end of: else if( MbQ->mbMode == AVC_I4 ...) */ } /* ***************************************************************************************** * \brief Filters one edge of 16 (luma) or 8 (chroma) pel ***************************************************************************************** */ void EdgeLoop_Luma_horizontal(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch) { int pel, ap = 0, aq = 0, Strng; int C0, c0, dif, AbsDelta, tmp, tmp1; int L2 = 0, L1, L0, R0, R1, R2 = 0, RL0; if (Strength[0] == 4) /* INTRA strong filtering */ { for (pel = 0; pel < 16; pel++) { R0 = SrcPtr[0]; R1 = SrcPtr[pitch]; L0 = SrcPtr[-pitch]; L1 = SrcPtr[-(pitch<<1)]; // |R0 - R1| < Beta tmp1 = R0 - R1; if (tmp1 < 0) tmp1 = -tmp1; tmp = (tmp1 - Beta); //|L0 - L1| < Beta tmp1 = L0 - L1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); //|R0 - L0| < Alpha AbsDelta = R0 - L0; if (AbsDelta < 0) AbsDelta = -AbsDelta; tmp &= (AbsDelta - Alpha); if (tmp < 0) { AbsDelta -= ((Alpha >> 2) + 2); R2 = SrcPtr[pitch<<1]; //inc2 L2 = SrcPtr[-(pitch+(pitch<<1))]; // -inc3 // |R0 - R2| < Beta && |R0 - L0| < (Alpha/4 + 2) tmp = R0 - R2; if (tmp < 0) tmp = -tmp; aq = AbsDelta & (tmp - Beta); // |L0 - L2| < Beta && |R0 - L0| < (Alpha/4 + 2) tmp = L0 - L2; if (tmp < 0) tmp = -tmp; ap = AbsDelta & (tmp - Beta); if (aq < 0) { tmp = R1 + R0 + L0; SrcPtr[0] = (L1 + (tmp << 1) + R2 + 4) >> 3; tmp += R2; SrcPtr[pitch] = (tmp + 2) >> 2; SrcPtr[pitch<<1] = (((SrcPtr[(pitch+(pitch<<1))] + R2) << 1) + tmp + 4) >> 3; } else SrcPtr[0] = ((R1 << 1) + R0 + L1 + 2) >> 2; if (ap < 0) { tmp = L1 + R0 + L0; SrcPtr[-pitch] = (R1 + (tmp << 1) + L2 + 4) >> 3; tmp += L2; SrcPtr[-(pitch<<1)] = (tmp + 2) >> 2; SrcPtr[-(pitch+(pitch<<1))] = (((SrcPtr[-(pitch<<2)] + L2) << 1) + tmp + 4) >> 3; } else SrcPtr[-pitch] = ((L1 << 1) + L0 + R1 + 2) >> 2; } /* if(tmp < 0) */ SrcPtr ++; // Increment to next set of pixel } /* end of: for(pel=0; pel<16; pel++) */ } /* if(Strength[0] == 4) */ else /* Normal filtering */ { for (pel = 0; pel < 16; pel++) { Strng = Strength[pel >> 2]; if (Strng) { R0 = SrcPtr[0]; R1 = SrcPtr[pitch]; L0 = SrcPtr[-pitch]; L1 = SrcPtr[-(pitch<<1)]; // inc2 //|R0 - L0| < Alpha tmp1 = R0 - L0; if (tmp1 < 0) tmp1 = -tmp1; tmp = (tmp1 - Alpha); // |R0 - R1| < Beta tmp1 = R0 - R1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); //|L0 - L1| < Beta tmp1 = L0 - L1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); if (tmp < 0) { R2 = SrcPtr[pitch<<1]; //inc2 L2 = SrcPtr[-(pitch+(pitch<<1))]; // -inc3 // |R0 - R2| < Beta tmp = R0 - R2; if (tmp < 0) tmp = -tmp; aq = tmp - Beta; // |L0 - L2| < Beta tmp = L0 - L2; if (tmp < 0) tmp = -tmp; ap = tmp - Beta; c0 = C0 = clipTable[Strng]; if (ap < 0) c0++; if (aq < 0) c0++; //dif = IClip(-c0, c0, ((Delta << 2) + (L1 - R1) + 4) >> 3); dif = (((R0 - L0) << 2) + (L1 - R1) + 4) >> 3; tmp = dif + c0; if ((uint)tmp > (uint)c0 << 1) { tmp = ~(tmp >> 31); dif = (tmp & (c0 << 1)) - c0; } //SrcPtr[0] = (uint8)IClip(0, 255, R0 - dif); //SrcPtr[-inc] = (uint8)IClip(0, 255, L0 + dif); RL0 = R0 + L0; R0 -= dif; L0 += dif; if ((uint)R0 > 255) { tmp = ~(R0 >> 31); R0 = tmp & 255; } if ((uint)L0 > 255) { tmp = ~(L0 >> 31); L0 = tmp & 255; } SrcPtr[-pitch] = L0; SrcPtr[0] = R0; if (C0 != 0) /* Multiple zeros in the clip tables */ { if (aq < 0) // SrcPtr[inc] += IClip(-C0, C0,(R2 + ((RL0 + 1) >> 1) - (R1<<1)) >> 1); { R2 = (R2 + ((RL0 + 1) >> 1) - (R1 << 1)) >> 1; tmp = R2 + C0; if ((uint)tmp > (uint)C0 << 1) { tmp = ~(tmp >> 31); R2 = (tmp & (C0 << 1)) - C0; } SrcPtr[pitch] += R2; } if (ap < 0) //SrcPtr[-inc2] += IClip(-C0, C0,(L2 + ((RL0 + 1) >> 1) - (L1<<1)) >> 1); { L2 = (L2 + ((RL0 + 1) >> 1) - (L1 << 1)) >> 1; tmp = L2 + C0; if ((uint)tmp > (uint)C0 << 1) { tmp = ~(tmp >> 31); L2 = (tmp & (C0 << 1)) - C0; } SrcPtr[-(pitch<<1)] += L2; } } } /* if(tmp < 0) */ } /* end of: if((Strng = Strength[pel >> 2])) */ SrcPtr ++; // Increment to next set of pixel } /* for(pel=0; pel<16; pel++) */ } /* else if(Strength[0] == 4) */ } void EdgeLoop_Luma_vertical(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch) { int pel, ap = 1, aq = 1; int C0, c0, dif, AbsDelta, Strng, tmp, tmp1; int L2 = 0, L1, L0, R0, R1, R2 = 0; uint8 *ptr, *ptr1; register uint R_in, L_in; uint R_out, L_out; if (Strength[0] == 4) /* INTRA strong filtering */ { for (pel = 0; pel < 16; pel++) { // Read 8 pels R_in = *((uint *)SrcPtr); // R_in = {R3, R2, R1, R0} L_in = *((uint *)(SrcPtr - 4)); // L_in = {L0, L1, L2, L3} R1 = (R_in >> 8) & 0xff; R0 = R_in & 0xff; L0 = L_in >> 24; L1 = (L_in >> 16) & 0xff; // |R0 - R1| < Beta tmp1 = (R_in & 0xff) - R1; if (tmp1 < 0) tmp1 = -tmp1; tmp = (tmp1 - Beta); //|L0 - L1| < Beta tmp1 = (L_in >> 24) - L1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); //|R0 - L0| < Alpha AbsDelta = (R_in & 0xff) - (L_in >> 24); if (AbsDelta < 0) AbsDelta = -AbsDelta; tmp &= (AbsDelta - Alpha); if (tmp < 0) { AbsDelta -= ((Alpha >> 2) + 2); R2 = (R_in >> 16) & 0xff; L2 = (L_in >> 8) & 0xff; // |R0 - R2| < Beta && |R0 - L0| < (Alpha/4 + 2) tmp1 = (R_in & 0xff) - R2; if (tmp1 < 0) tmp1 = -tmp1; aq = AbsDelta & (tmp1 - Beta); // |L0 - L2| < Beta && |R0 - L0| < (Alpha/4 + 2) tmp1 = (L_in >> 24) - L2; if (tmp1 < 0) tmp1 = -tmp1; ap = AbsDelta & (tmp1 - Beta); ptr = SrcPtr; if (aq < 0) { R_out = (R_in >> 24) << 24; // Keep R3 at the fourth byte tmp = R0 + L0 + R1; R_out |= (((tmp << 1) + L1 + R2 + 4) >> 3); tmp += R2; R_out |= (((tmp + 2) >> 2) << 8); tmp1 = ((R_in >> 24) + R2) << 1; R_out |= (((tmp1 + tmp + 4) >> 3) << 16); *((uint *)SrcPtr) = R_out; } else *ptr = ((R1 << 1) + R0 + L1 + 2) >> 2; if (ap < 0) { L_out = (L_in << 24) >> 24; // Keep L3 at the first byte tmp = R0 + L0 + L1; L_out |= ((((tmp << 1) + R1 + L2 + 4) >> 3) << 24); tmp += L2; L_out |= (((tmp + 2) >> 2) << 16); tmp1 = ((L_in & 0xff) + L2) << 1; L_out |= (((tmp1 + tmp + 4) >> 3) << 8); *((uint *)(SrcPtr - 4)) = L_out; } else *(--ptr) = ((L1 << 1) + L0 + R1 + 2) >> 2; } /* if(tmp < 0) */ SrcPtr += pitch; // Increment to next set of pixel } /* end of: for(pel=0; pel<16; pel++) */ } /* if(Strength[0] == 4) */ else /* Normal filtering */ { for (pel = 0; pel < 16; pel++) { Strng = Strength[pel >> 2]; if (Strng) { // Read 8 pels R_in = *((uint *)SrcPtr); // R_in = {R3, R2, R1, R0} L_in = *((uint *)(SrcPtr - 4)); // L_in = {L0, L1, L2, L3} R1 = (R_in >> 8) & 0xff; R0 = R_in & 0xff; L0 = L_in >> 24; L1 = (L_in >> 16) & 0xff; //|R0 - L0| < Alpha tmp = R0 - L0; if (tmp < 0) tmp = -tmp; tmp -= Alpha; // |R0 - R1| < Beta tmp1 = R0 - R1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); //|L0 - L1| < Beta tmp1 = L0 - L1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); if (tmp < 0) { L2 = SrcPtr[-3]; R2 = SrcPtr[2]; // |R0 - R2| < Beta tmp = R0 - R2; if (tmp < 0) tmp = -tmp; aq = tmp - Beta; // |L0 - L2| < Beta tmp = L0 - L2; if (tmp < 0) tmp = -tmp; ap = tmp - Beta; c0 = C0 = clipTable[Strng]; if (ap < 0) c0++; if (aq < 0) c0++; //dif = IClip(-c0, c0, ((Delta << 2) + (L1 - R1) + 4) >> 3); dif = (((R0 - L0) << 2) + (L1 - R1) + 4) >> 3; tmp = dif + c0; if ((uint)tmp > (uint)c0 << 1) { tmp = ~(tmp >> 31); dif = (tmp & (c0 << 1)) - c0; } ptr = SrcPtr; ptr1 = SrcPtr - 1; //SrcPtr[0] = (uint8)IClip(0, 255, R0 - dif); //SrcPtr[-inc] = (uint8)IClip(0, 255, L0 + dif); R_in = R0 - dif; L_in = L0 + dif; /* cannot re-use R0 and L0 here */ if ((uint)R_in > 255) { tmp = ~((int)R_in >> 31); R_in = tmp & 255; } if ((uint)L_in > 255) { tmp = ~((int)L_in >> 31); L_in = tmp & 255; } *ptr1-- = L_in; *ptr++ = R_in; if (C0 != 0) // Multiple zeros in the clip tables { if (ap < 0) //SrcPtr[-inc2] += IClip(-C0, C0,(L2 + ((RL0 + 1) >> 1) - (L1<<1)) >> 1); { L2 = (L2 + ((R0 + L0 + 1) >> 1) - (L1 << 1)) >> 1; tmp = L2 + C0; if ((uint)tmp > (uint)C0 << 1) { tmp = ~(tmp >> 31); L2 = (tmp & (C0 << 1)) - C0; } *ptr1 += L2; } if (aq < 0) // SrcPtr[inc] += IClip(-C0, C0,(R2 + ((RL0 + 1) >> 1) - (R1<<1)) >> 1); { R2 = (R2 + ((R0 + L0 + 1) >> 1) - (R1 << 1)) >> 1; tmp = R2 + C0; if ((uint)tmp > (uint)C0 << 1) { tmp = ~(tmp >> 31); R2 = (tmp & (C0 << 1)) - C0; } *ptr += R2; } } } /* if(tmp < 0) */ } /* end of: if((Strng = Strength[pel >> 2])) */ SrcPtr += pitch; // Increment to next set of pixel } /* for(pel=0; pel<16; pel++) */ } /* else if(Strength[0] == 4) */ } void EdgeLoop_Chroma_vertical(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch) { int pel, Strng; int c0, dif; int L1, L0, R0, R1, tmp, tmp1; uint8 *ptr; uint R_in, L_in; for (pel = 0; pel < 16; pel++) { Strng = Strength[pel>>2]; if (Strng) { // Read 8 pels R_in = *((uint *)SrcPtr); // R_in = {R3, R2, R1, R0} L_in = *((uint *)(SrcPtr - 4)); // L_in = {L0, L1, L2, L3} R1 = (R_in >> 8) & 0xff; R0 = R_in & 0xff; L0 = L_in >> 24; L1 = (L_in >> 16) & 0xff; // |R0 - R1| < Beta tmp1 = R0 - R1; if (tmp1 < 0) tmp1 = -tmp1; tmp = (tmp1 - Beta); //|L0 - L1| < Beta tmp1 = L0 - L1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); //|R0 - L0| < Alpha tmp1 = R0 - L0; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Alpha); if (tmp < 0) { ptr = SrcPtr; if (Strng == 4) /* INTRA strong filtering */ { *ptr-- = ((R1 << 1) + R0 + L1 + 2) >> 2; *ptr = ((L1 << 1) + L0 + R1 + 2) >> 2; } else /* normal filtering */ { c0 = clipTable[Strng] + 1; //dif = IClip(-c0, c0, ((Delta << 2) + (L1 - R1) + 4) >> 3); dif = (((R0 - L0) << 2) + (L1 - R1) + 4) >> 3; tmp = dif + c0; if ((uint)tmp > (uint)c0 << 1) { tmp = ~(tmp >> 31); dif = (tmp & (c0 << 1)) - c0; } //SrcPtr[0] = (uint8)IClip(0, 255, R0 - dif); //SrcPtr[-inc] = (uint8)IClip(0, 255, L0 + dif); L0 += dif; R0 -= dif; if ((uint)L0 > 255) { tmp = ~(L0 >> 31); L0 = tmp & 255; } if ((uint)R0 > 255) { tmp = ~(R0 >> 31); R0 = tmp & 255; } *ptr-- = R0; *ptr = L0; } } pel ++; SrcPtr += pitch; // Increment to next set of pixel } /* end of: if((Strng = Strength[pel >> 2])) */ else { pel += 3; SrcPtr += (pitch << 1); //PtrInc << 1; } } /* end of: for(pel=0; pel<16; pel++) */ } void EdgeLoop_Chroma_horizontal(uint8* SrcPtr, uint8 *Strength, int Alpha, int Beta, int *clipTable, int pitch) { int pel, Strng; int c0, dif; int L1, L0, R0, R1, tmp, tmp1; for (pel = 0; pel < 16; pel++) { Strng = Strength[pel>>2]; if (Strng) { R0 = SrcPtr[0]; L0 = SrcPtr[-pitch]; L1 = SrcPtr[-(pitch<<1)]; //inc2 R1 = SrcPtr[pitch]; // |R0 - R1| < Beta tmp1 = R0 - R1; if (tmp1 < 0) tmp1 = -tmp1; tmp = (tmp1 - Beta); //|L0 - L1| < Beta tmp1 = L0 - L1; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Beta); //|R0 - L0| < Alpha tmp1 = R0 - L0; if (tmp1 < 0) tmp1 = -tmp1; tmp &= (tmp1 - Alpha); if (tmp < 0) { if (Strng == 4) /* INTRA strong filtering */ { SrcPtr[0] = ((R1 << 1) + R0 + L1 + 2) >> 2; SrcPtr[-pitch] = ((L1 << 1) + L0 + R1 + 2) >> 2; } else /* normal filtering */ { c0 = clipTable[Strng] + 1; //dif = IClip(-c0, c0, ((Delta << 2) + (L1 - R1) + 4) >> 3); dif = (((R0 - L0) << 2) + (L1 - R1) + 4) >> 3; tmp = dif + c0; if ((uint)tmp > (uint)c0 << 1) { tmp = ~(tmp >> 31); dif = (tmp & (c0 << 1)) - c0; } //SrcPtr[-inc] = (uint8)IClip(0, 255, L0 + dif); //SrcPtr[0] = (uint8)IClip(0, 255, R0 - dif); L0 += dif; R0 -= dif; if ((uint)L0 > 255) { tmp = ~(L0 >> 31); L0 = tmp & 255; } if ((uint)R0 > 255) { tmp = ~(R0 >> 31); R0 = tmp & 255; } SrcPtr[0] = R0; SrcPtr[-pitch] = L0; } } pel ++; SrcPtr ++; // Increment to next set of pixel } /* end of: if((Strng = Strength[pel >> 2])) */ else { pel += 3; SrcPtr += 2; } } /* end of: for(pel=0; pel<16; pel++) */ }