HEVC编码技术解析与应用
本文深入探讨了HEVC编码技术的核心原理与实践应用，包括几何信息计算、CU数据初始化、熵编码参数获取等内容，重点阐述了尺寸小的TU（Luma8x8/4x4,Chroma4x4/2x2）的扫描方式选择以及预测方向对扫描顺序的影响。文章旨在为理解和优化HEVC编码提供技术支持。
注：问号以及未注释部分 会在x265-1.9版本内更新
/*****************************************************************************
* Copyright (C) 2015 x265 project
*
* Authors: Steve Borho <steve@borho.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
*
* This program is also available under a commercial proprietary license.
* For more information, contact us at license @ x265.com.
*****************************************************************************/
#include "common.h"
#include "frame.h"
#include "framedata.h"
#include "picyuv.h"
#include "mv.h"
#include "cudata.h"
using namespace X265_NS;
/* for all bcast* and copy* functions, dst and src are aligned to MIN(size, 32) */
static void bcast1(uint8_t* dst, uint8_t val) { dst[0] = val; }//一次性set 1个数据
static void copy4(uint8_t* dst, uint8_t* src) { ((uint32_t*)dst)[0] = ((uint32_t*)src)[0]; }//一次性copy 4个数据 分成1个32位数字赋值
static void bcast4(uint8_t* dst, uint8_t val) { ((uint32_t*)dst)[0] = 0x01010101u * val; }//一次性set 4个数据 分成1个32为数字赋值
static void copy16(uint8_t* dst, uint8_t* src) { ((uint64_t*)dst)[0] = ((uint64_t*)src)[0]; ((uint64_t*)dst)[1] = ((uint64_t*)src)[1]; }//一次性copy 16个数据 分成2个64位数字赋值
static void bcast16(uint8_t* dst, uint8_t val) { uint64_t bval = 0x0101010101010101ULL * val; ((uint64_t*)dst)[0] = bval; ((uint64_t*)dst)[1] = bval; }//一次性set 16个数据 分成2个64位数字赋值
static void copy64(uint8_t* dst, uint8_t* src) { ((uint64_t*)dst)[0] = ((uint64_t*)src)[0]; ((uint64_t*)dst)[1] = ((uint64_t*)src)[1];
((uint64_t*)dst)[2] = ((uint64_t*)src)[2]; ((uint64_t*)dst)[3] = ((uint64_t*)src)[3];
((uint64_t*)dst)[4] = ((uint64_t*)src)[4]; ((uint64_t*)dst)[5] = ((uint64_t*)src)[5];
((uint64_t*)dst)[6] = ((uint64_t*)src)[6]; ((uint64_t*)dst)[7] = ((uint64_t*)src)[7]; }//一次性copy 64个数据 分成8个64位数字赋值
static void bcast64(uint8_t* dst, uint8_t val) { uint64_t bval = 0x0101010101010101ULL * val;
((uint64_t*)dst)[0] = bval; ((uint64_t*)dst)[1] = bval; ((uint64_t*)dst)[2] = bval; ((uint64_t*)dst)[3] = bval;
((uint64_t*)dst)[4] = bval; ((uint64_t*)dst)[5] = bval; ((uint64_t*)dst)[6] = bval; ((uint64_t*)dst)[7] = bval; }//一次性set 64个数据 分成8个64位数字赋值
/* at 256 bytes, memset/memcpy will probably use SIMD more effectively than our uint64_t hack,
* but hand-written assembly would beat it. */
static void copy256(uint8_t* dst, uint8_t* src) { memcpy(dst, src, 256); }//一次性copy256个数据
static void bcast256(uint8_t* dst, uint8_t val) { memset(dst, val, 256); }//一次性set256个数据
namespace {
// file private namespace
/* Check whether 2 addresses point to the same column */
/** 函数功能 ： 判断addrA、addrB是否处于同一列
* \参数 addrA ： PU左上角像素点在CTU中的光栅扫描位置
* \参数 addrB ： 当前CU在CTU的光栅扫描位置
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回addrA、addrB是否处于同一列**/
inline bool isEqualCol(int addrA, int addrB, int numUnits)
{
// addrA % numUnits == addrB % numUnits
return ((addrA ^ addrB) & (numUnits - 1)) == 0;
}
/* Check whether 2 addresses point to the same row */
/** 函数功能 ： 判断addrA、addrB是否处于同一行
* \参数 addrA ： PU左上角像素点在CTU中的光栅扫描位置
* \参数 addrB ： 当前CU在CTU的光栅扫描位置
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回addrA、addrB是否处于同一行**/
inline bool isEqualRow(int addrA, int addrB, int numUnits)
{
// addrA / numUnits == addrB / numUnits
return ((addrA ^ addrB) & ~(numUnits - 1)) == 0;
}
/* Check whether 2 addresses point to the same row or column */
/** 函数功能 ： 返回addrA、addrB是否处于同一行或者同一列
* \参数 addrA ： PU左上角像素点在CTU中的光栅扫描位置
* \参数 addrB ： 当前CU在CTU的光栅扫描位置
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回addrA、addrB是否处于同一行或者同一列**/
inline bool isEqualRowOrCol(int addrA, int addrB, int numUnits)
{
return isEqualCol(addrA, addrB, numUnits) | isEqualRow(addrA, addrB, numUnits);
}
/* Check whether one address points to the first column */
/** 函数功能 ： 返回当前位置是否处在CTU列第一列中（是返回ture 否返回false）
* \参数 addr ： 当前位置在CTU的光栅扫描位置
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回当前位置是否处在CTU列第一列中**/
inline bool isZeroCol(int addr, int numUnits)
{
// addr % numUnits == 0
return (addr & (numUnits - 1)) == 0;//判断是否为第一列
}
/* Check whether one address points to the first row */
/** 函数功能 ： 返回当前位置是否处在CTU行第一行中（是返回ture 否返回false）
* \参数 addr ： 当前位置在CTU的光栅扫描位置
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回当前位置是否处在CTU列第一行中**/
inline bool isZeroRow(int addr, int numUnits)
{
// addr / numUnits == 0
return (addr & ~(numUnits - 1)) == 0;//判断是否是第一行
}
/* Check whether one address points to a column whose index is smaller than a given value */
/** 函数功能 ： 返回当前参考块是否在同一个CTU内部 针对列
* \参数 addr ： 当前PU位置在CTU的光栅扫描位置
* \参数 val ： numUnits - 当前计数的4x4块
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回当前参考块是否在同一个CTU内部**/
inline bool lessThanCol(int addr, int val, int numUnits)
{
// addr % numUnits < val
return (addr & (numUnits - 1)) < val;//PU右上角位置在X中的坐标 小于 numUnits - 当前计数的4x4块
}
/** 函数功能 ： 返回当前参考块是否在同一个CTU内部 针对行
* \参数 addr ： 当前PU位置在CTU的光栅扫描位置
* \参数 val ： numUnits - 当前计数的4x4块
* \参数 numUnits ： 一个CTU中每行或者列有多少4x4块
* 返回值 ： 返回当前参考块是否在同一个CTU内部**/
/* Check whether one address points to a row whose index is smaller than a given value */
inline bool lessThanRow(int addr, int val, int numUnits)
{
// addr / numUnits < val
return addr < val * numUnits;
}
inline MV scaleMv(MV mv, int scale)
{
int mvx = x265_clip3(-32768, 32767, (scale * mv.x + 127 + (scale * mv.x < 0)) >> 8);
int mvy = x265_clip3(-32768, 32767, (scale * mv.y + 127 + (scale * mv.y < 0)) >> 8);
return MV((int16_t)mvx, (int16_t)mvy);
}
}
cubcast_t CUData::s_partSet[NUM_FULL_DEPTH] = { NULL, NULL, NULL, NULL, NULL };//set 函数指针
uint32_t CUData::s_numPartInCUSize;//存储一个CTU中每行或者列有多少4x4块
CUData::CUData()//初始化
{
memset(this, 0, sizeof(*this));
}
/** 函数功能 ：初始化函数指针，获取CTU/CU数据相应存储位置
/* 调用范围 ：只在FrameData::create和Analysis::create函数中被调用
* \参数 dataPool ：CU存储空间
* \参数 depth ：CU划分深度（FrameData::create 传入 0 Analysis::create 传入相应深度）
* \参数 csp ：图像格式
* \参数 instance ：每个CU的标号
* \返回 ：null* */
void CUData::initialize(const CUDataMemPool& dataPool, uint32_t depth, int csp, int instance)
{
m_chromaFormat = csp; //获取图像格式
m_hChromaShift = CHROMA_H_SHIFT(csp);//获取移位个数
m_vChromaShift = CHROMA_V_SHIFT(csp);//获取移位个数
m_numPartitions = NUM_4x4_PARTITIONS >> (depth * 2);//获取当前CU的4x4块个数
if (!s_partSet[0])
{
s_numPartInCUSize = 1 << g_unitSizeDepth;//获取一个CTU中每行或者列有多少4x4块
switch (g_maxLog2CUSize) //根据CTU大小 配置相应set函数指针 （所有CUdata类中 只配置一次 共用）
{
case 6:
s_partSet[0] = bcast256;
s_partSet[1] = bcast64;
s_partSet[2] = bcast16;
s_partSet[3] = bcast4;
s_partSet[4] = bcast1;
break;
case 5:
s_partSet[0] = bcast64;
s_partSet[1] = bcast16;
s_partSet[2] = bcast4;
s_partSet[3] = bcast1;
s_partSet[4] = NULL;
break;
case 4:
s_partSet[0] = bcast16;
s_partSet[1] = bcast4;
s_partSet[2] = bcast1;
s_partSet[3] = NULL;
s_partSet[4] = NULL;
break;
default:
X265_CHECK(0, "unexpected CTU size\n");
break;
}
}
switch (m_numPartitions)//根据具体的CU 4x4块个数选择具体的copy、set、subcopy、subset函数指针
{
case 256: // 64x64 CU
m_partCopy = copy256;
m_partSet = bcast256;
m_subPartCopy = copy64;
m_subPartSet = bcast64;
break;
case 64: // 32x32 CU
m_partCopy = copy64;
m_partSet = bcast64;
m_subPartCopy = copy16;
m_subPartSet = bcast16;
break;
case 16: // 16x16 CU
m_partCopy = copy16;
m_partSet = bcast16;
m_subPartCopy = copy4;
m_subPartSet = bcast4;
break;
case 4: // 8x8 CU
m_partCopy = copy4;
m_partSet = bcast4;
m_subPartCopy = NULL;
m_subPartSet = NULL;
break;
default:
X265_CHECK(0, "unexpected CU partition count\n");
break;
}
/* Each CU's data is layed out sequentially within the charMemBlock */
uint8_t *charBuf = dataPool.charMemBlock + (m_numPartitions * BytesPerPartition) * instance;//在dataPool中获取当前CTU标号的首地址
m_qp = (int8_t*)charBuf; charBuf += m_numPartitions;//以下获取相应地址
m_log2CUSize = charBuf; charBuf += m_numPartitions;
m_lumaIntraDir = charBuf; charBuf += m_numPartitions;
m_tqBypass = charBuf; charBuf += m_numPartitions;
m_refIdx[0] = (int8_t*)charBuf; charBuf += m_numPartitions;
m_refIdx[1] = (int8_t*)charBuf; charBuf += m_numPartitions;
m_cuDepth = charBuf; charBuf += m_numPartitions;
m_predMode = charBuf; charBuf += m_numPartitions; /* the order up to here is important in initCTU() and initSubCU() */
m_partSize = charBuf; charBuf += m_numPartitions;
m_mergeFlag = charBuf; charBuf += m_numPartitions;
m_interDir = charBuf; charBuf += m_numPartitions;
m_mvpIdx[0] = charBuf; charBuf += m_numPartitions;
m_mvpIdx[1] = charBuf; charBuf += m_numPartitions;
m_tuDepth = charBuf; charBuf += m_numPartitions;
m_transformSkip[0] = charBuf; charBuf += m_numPartitions;
m_transformSkip[1] = charBuf; charBuf += m_numPartitions;
m_transformSkip[2] = charBuf; charBuf += m_numPartitions;
m_cbf[0] = charBuf; charBuf += m_numPartitions;
m_cbf[1] = charBuf; charBuf += m_numPartitions;
m_cbf[2] = charBuf; charBuf += m_numPartitions;
m_chromaIntraDir = charBuf; charBuf += m_numPartitions;
X265_CHECK(charBuf == dataPool.charMemBlock + (m_numPartitions * BytesPerPartition) * (instance + 1), "CU data layout is broken\n");
m_mv[0] = dataPool.mvMemBlock + (instance * 4) * m_numPartitions;//获取MV等相应地址
m_mv[1] = m_mv[0] + m_numPartitions;
m_mvd[0] = m_mv[1] + m_numPartitions;
m_mvd[1] = m_mvd[0] + m_numPartitions;
uint32_t cuSize = g_maxCUSize >> depth;
uint32_t sizeL = cuSize * cuSize;
uint32_t sizeC = sizeL >> (m_hChromaShift + m_vChromaShift);
m_trCoeff[0] = dataPool.trCoeffMemBlock + instance * (sizeL + sizeC * 2);//获取残差系数相应地址
m_trCoeff[1] = m_trCoeff[0] + sizeL;
m_trCoeff[2] = m_trCoeff[0] + sizeL + sizeC;
}
/** 函数功能 ：初始化CTU
/* 调用范围 ：只在processRowEncoder函数中被调用
* \参数 frame ：当前编码帧
* \参数 cuAddr ：CTU在帧中的编号
* \参数 qp ：通过ratecontrolstart估计的QP值
* \返回 ：null* */
void CUData::initCTU(const Frame& frame, uint32_t cuAddr, int qp)
{
m_encData = frame.m_encData;//获取当前帧编码数据
m_slice = m_encData->m_slice;
m_cuAddr = cuAddr;
m_cuPelX = (cuAddr % m_slice->m_sps->numCuInWidth) << g_maxLog2CUSize;
m_cuPelY = (cuAddr / m_slice->m_sps->numCuInWidth) << g_maxLog2CUSize;
m_absIdxInCTU = 0;
m_numPartitions = NUM_4x4_PARTITIONS;
/* sequential memsets */
m_partSet((uint8_t*)m_qp, (uint8_t)qp);
m_partSet(m_log2CUSize, (uint8_t)g_maxLog2CUSize);
m_partSet(m_lumaIntraDir, (uint8_t)DC_IDX);
m_partSet(m_tqBypass, (uint8_t)frame.m_encData->m_param->bLossless);
if (m_slice->m_sliceType != I_SLICE)
{
m_partSet((uint8_t*)m_refIdx[0], (uint8_t)REF_NOT_VALID);
m_partSet((uint8_t*)m_refIdx[1], (uint8_t)REF_NOT_VALID);
}
X265_CHECK(!(frame.m_encData->m_param->bLossless && !m_slice->m_pps->bTransquantBypassEnabled), "lossless enabled without TQbypass in PPS\n");
/* initialize the remaining CU data in one memset */
memset(m_cuDepth, 0, (BytesPerPartition - 6) * m_numPartitions);
uint32_t widthInCU = m_slice->m_sps->numCuInWidth;
m_cuLeft = (m_cuAddr % widthInCU) ? m_encData->getPicCTU(m_cuAddr - 1) : NULL; //获取左边CTU
m_cuAbove = (m_cuAddr / widthInCU) ? m_encData->getPicCTU(m_cuAddr - widthInCU) : NULL;//获取上边CTU
m_cuAboveLeft = (m_cuLeft && m_cuAbove) ? m_encData->getPicCTU(m_cuAddr - widthInCU - 1) : NULL;//获取左上CTU
m_cuAboveRight = (m_cuAbove && ((m_cuAddr % widthInCU) < (widthInCU - 1))) ? m_encData->getPicCTU(m_cuAddr - widthInCU + 1) : NULL;//获取右上CTU
}
// initialize Sub partition
/** 函数功能 ： 根据cuGeom初始化当前子块的信息
* \参数 ctu ： 上层大块
* \参数 cuGeom ： 当前子块对应的一些几何（深度，位置等信息）信息
* \参数 qp ： 当前的qp信息
* 返回值 ： null**/
void CUData::initSubCU(const CUData& ctu, const CUGeom& cuGeom, int qp)
{
m_absIdxInCTU = cuGeom.absPartIdx;//获取子CU左上角在CTU4x4块中的zigzag标号
m_encData = ctu.m_encData; //获取编码数据
m_slice = ctu.m_slice; //获取slice
m_cuAddr = ctu.m_cuAddr; //获取CTU在帧中的编号
m_cuPelX = ctu.m_cuPelX + g_zscanToPelX[cuGeom.absPartIdx];//获取子CU左上角像素点在图像中的横向偏移像素个数
m_cuPelY = ctu.m_cuPelY + g_zscanToPelY[cuGeom.absPartIdx];//获取子CU左上角像素点在图像中的纵向偏移像素个数
m_cuLeft = ctu.m_cuLeft; //获取左边CTU
m_cuAbove = ctu.m_cuAbove;//获取上边CTU
m_cuAboveLeft = ctu.m_cuAboveLeft;//获取左上CTU
m_cuAboveRight = ctu.m_cuAboveRight;//获取右上CTU
X265_CHECK(m_numPartitions == cuGeom.numPartitions, "initSubCU() size mismatch\n");
m_partSet((uint8_t*)m_qp, (uint8_t)qp);//获取qp
m_partSet(m_log2CUSize, (uint8_t)cuGeom.log2CUSize);//获取当前块的块大小
m_partSet(m_lumaIntraDir, (uint8_t)DC_IDX);//初始化预测方向为DC
m_partSet(m_tqBypass, (uint8_t)m_encData->m_param->bLossless);//初始化跳过模式
m_partSet((uint8_t*)m_refIdx[0], (uint8_t)REF_NOT_VALID);//初始化最优参考帧标号
m_partSet((uint8_t*)m_refIdx[1], (uint8_t)REF_NOT_VALID);//初始化最优参考帧标号
m_partSet(m_cuDepth, (uint8_t)cuGeom.depth);//初始化当前块的深度
/* initialize the remaining CU data in one memset */
memset(m_predMode, 0, (BytesPerPartition - 7) * m_numPartitions);//剩下的所有相关信息全部初始化为0
}
/* Copy the results of a sub-part (split) CU to the parent CU */
/** 函数功能 ： 将子块决策最优结果copy到父块对应位置中
* \参数 subCU ： 当前子块
* \参数 childGeom ： 当前子块的几何信息
* \参数 subPartIdx ： 当前子块的标号
* 返回值 ： null**/
void CUData::copyPartFrom(const CUData& subCU, const CUGeom& childGeom, uint32_t subPartIdx)
{
X265_CHECK(subPartIdx < 4, "part unit should be less than 4\n");
uint32_t offset = childGeom.numPartitions * subPartIdx;//获取当前子块相对于父块的偏移4x4块个数
m_subPartCopy((uint8_t*)m_qp + offset, (uint8_t*)subCU.m_qp);
m_subPartCopy(m_log2CUSize + offset, subCU.m_log2CUSize);
m_subPartCopy(m_lumaIntraDir + offset, subCU.m_lumaIntraDir);
m_subPartCopy(m_tqBypass + offset, subCU.m_tqBypass);
m_subPartCopy((uint8_t*)m_refIdx[0] + offset, (uint8_t*)subCU.m_refIdx[0]);
m_subPartCopy((uint8_t*)m_refIdx[1] + offset, (uint8_t*)subCU.m_refIdx[1]);
m_subPartCopy(m_cuDepth + offset, subCU.m_cuDepth);
m_subPartCopy(m_predMode + offset, subCU.m_predMode);
m_subPartCopy(m_partSize + offset, subCU.m_partSize);
m_subPartCopy(m_mergeFlag + offset, subCU.m_mergeFlag);
m_subPartCopy(m_interDir + offset, subCU.m_interDir);
m_subPartCopy(m_mvpIdx[0] + offset, subCU.m_mvpIdx[0]);
m_subPartCopy(m_mvpIdx[1] + offset, subCU.m_mvpIdx[1]);
m_subPartCopy(m_tuDepth + offset, subCU.m_tuDepth);
m_subPartCopy(m_transformSkip[0] + offset, subCU.m_transformSkip[0]);
m_subPartCopy(m_transformSkip[1] + offset, subCU.m_transformSkip[1]);
m_subPartCopy(m_transformSkip[2] + offset, subCU.m_transformSkip[2]);
m_subPartCopy(m_cbf[0] + offset, subCU.m_cbf[0]);
m_subPartCopy(m_cbf[1] + offset, subCU.m_cbf[1]);
m_subPartCopy(m_cbf[2] + offset, subCU.m_cbf[2]);
m_subPartCopy(m_chromaIntraDir + offset, subCU.m_chromaIntraDir);
memcpy(m_mv[0] + offset, subCU.m_mv[0], childGeom.numPartitions * sizeof(MV));
memcpy(m_mv[1] + offset, subCU.m_mv[1], childGeom.numPartitions * sizeof(MV));
memcpy(m_mvd[0] + offset, subCU.m_mvd[0], childGeom.numPartitions * sizeof(MV));
memcpy(m_mvd[1] + offset, subCU.m_mvd[1], childGeom.numPartitions * sizeof(MV));
uint32_t tmp = 1 << ((g_maxLog2CUSize - childGeom.depth) * 2);
uint32_t tmp2 = subPartIdx * tmp;
memcpy(m_trCoeff[0] + tmp2, subCU.m_trCoeff[0], sizeof(coeff_t) * tmp);
uint32_t tmpC = tmp >> (m_hChromaShift + m_vChromaShift);
uint32_t tmpC2 = tmp2 >> (m_hChromaShift + m_vChromaShift);
memcpy(m_trCoeff[1] + tmpC2, subCU.m_trCoeff[1], sizeof(coeff_t) * tmpC);
memcpy(m_trCoeff[2] + tmpC2, subCU.m_trCoeff[2], sizeof(coeff_t) * tmpC);
}
/* If a sub-CU part is not present (off the edge of the picture) its depth and
* log2size should still be configured */
/** 函数功能 ： 当前子块全部在图像外，设置当前子块的depth和块大小值
* \参数 childGeom ： 当前子块的几何信息
* \参数 subPartIdx ： 当前子块的标号
* 返回值 ： null**/
void CUData::setEmptyPart(const CUGeom& childGeom, uint32_t subPartIdx)
{
uint32_t offset = childGeom.numPartitions * subPartIdx; //在当前子块与当前父块之间的偏移4x4块个数
m_subPartSet(m_cuDepth + offset, (uint8_t)childGeom.depth); //赋值当前子块的depth
m_subPartSet(m_log2CUSize + offset, (uint8_t)childGeom.log2CUSize);//赋值当前子块的cu大小
}
/* Copy all CU data from one instance to the next, except set lossless flag
* This will only get used when --cu-lossless is enabled but --lossless is not. */
void CUData::initLosslessCU(const CUData& cu, const CUGeom& cuGeom)
{
/* Start by making an exact copy */
m_encData = cu.m_encData;
m_slice = cu.m_slice;
m_cuAddr = cu.m_cuAddr;
m_cuPelX = cu.m_cuPelX;
m_cuPelY = cu.m_cuPelY;
m_cuLeft = cu.m_cuLeft;
m_cuAbove = cu.m_cuAbove;
m_cuAboveLeft = cu.m_cuAboveLeft;
m_cuAboveRight = cu.m_cuAboveRight;
m_absIdxInCTU = cuGeom.absPartIdx;
m_numPartitions = cuGeom.numPartitions;
memcpy(m_qp, cu.m_qp, BytesPerPartition * m_numPartitions);
memcpy(m_mv[0], cu.m_mv[0], m_numPartitions * sizeof(MV));
memcpy(m_mv[1], cu.m_mv[1], m_numPartitions * sizeof(MV));
memcpy(m_mvd[0], cu.m_mvd[0], m_numPartitions * sizeof(MV));
memcpy(m_mvd[1], cu.m_mvd[1], m_numPartitions * sizeof(MV));
/* force TQBypass to true */
m_partSet(m_tqBypass, true);
/* clear residual coding flags */
m_partSet(m_predMode, cu.m_predMode[0] & (MODE_INTRA | MODE_INTER));
m_partSet(m_tuDepth, 0);
m_partSet(m_transformSkip[0], 0);
m_partSet(m_transformSkip[1], 0);
m_partSet(m_transformSkip[2], 0);
m_partSet(m_cbf[0], 0);
m_partSet(m_cbf[1], 0);
m_partSet(m_cbf[2], 0);
}
/* Copy completed predicted CU to CTU in picture */
/** 函数功能 ： 将当前cu决策信息赋值到CTU对应位置中
* \参数 depth ： 当前的划分深度
* 返回值 ： null**/
void CUData::copyToPic(uint32_t depth) const
{
CUData& ctu = *m_encData->getPicCTU(m_cuAddr);//获取当前CU所在的CTU
//将当前cu决策信息赋值到CTU对应位置中
m_partCopy((uint8_t*)ctu.m_qp + m_absIdxInCTU, (uint8_t*)m_qp);
m_partCopy(ctu.m_log2CUSize + m_absIdxInCTU, m_log2CUSize);
m_partCopy(ctu.m_lumaIntraDir + m_absIdxInCTU, m_lumaIntraDir);
m_partCopy(ctu.m_tqBypass + m_absIdxInCTU, m_tqBypass);
m_partCopy((uint8_t*)ctu.m_refIdx[0] + m_absIdxInCTU, (uint8_t*)m_refIdx[0]);
m_partCopy((uint8_t*)ctu.m_refIdx[1] + m_absIdxInCTU, (uint8_t*)m_refIdx[1]);
m_partCopy(ctu.m_cuDepth + m_absIdxInCTU, m_cuDepth);
m_partCopy(ctu.m_predMode + m_absIdxInCTU, m_predMode);
m_partCopy(ctu.m_partSize + m_absIdxInCTU, m_partSize);
m_partCopy(ctu.m_mergeFlag + m_absIdxInCTU, m_mergeFlag);
m_partCopy(ctu.m_interDir + m_absIdxInCTU, m_interDir);
m_partCopy(ctu.m_mvpIdx[0] + m_absIdxInCTU, m_mvpIdx[0]);
m_partCopy(ctu.m_mvpIdx[1] + m_absIdxInCTU, m_mvpIdx[1]);
m_partCopy(ctu.m_tuDepth + m_absIdxInCTU, m_tuDepth);
m_partCopy(ctu.m_transformSkip[0] + m_absIdxInCTU, m_transformSkip[0]);
m_partCopy(ctu.m_transformSkip[1] + m_absIdxInCTU, m_transformSkip[1]);
m_partCopy(ctu.m_transformSkip[2] + m_absIdxInCTU, m_transformSkip[2]);
m_partCopy(ctu.m_cbf[0] + m_absIdxInCTU, m_cbf[0]);
m_partCopy(ctu.m_cbf[1] + m_absIdxInCTU, m_cbf[1]);
m_partCopy(ctu.m_cbf[2] + m_absIdxInCTU, m_cbf[2]);
m_partCopy(ctu.m_chromaIntraDir + m_absIdxInCTU, m_chromaIntraDir);
memcpy(ctu.m_mv[0] + m_absIdxInCTU, m_mv[0], m_numPartitions * sizeof(MV));
memcpy(ctu.m_mv[1] + m_absIdxInCTU, m_mv[1], m_numPartitions * sizeof(MV));
memcpy(ctu.m_mvd[0] + m_absIdxInCTU, m_mvd[0], m_numPartitions * sizeof(MV));
memcpy(ctu.m_mvd[1] + m_absIdxInCTU, m_mvd[1], m_numPartitions * sizeof(MV));
uint32_t tmpY = 1 << ((g_maxLog2CUSize - depth) * 2);
uint32_t tmpY2 = m_absIdxInCTU << (LOG2_UNIT_SIZE * 2);
memcpy(ctu.m_trCoeff[0] + tmpY2, m_trCoeff[0], sizeof(coeff_t) * tmpY);
uint32_t tmpC = tmpY >> (m_hChromaShift + m_vChromaShift);
uint32_t tmpC2 = tmpY2 >> (m_hChromaShift + m_vChromaShift);
memcpy(ctu.m_trCoeff[1] + tmpC2, m_trCoeff[1], sizeof(coeff_t) * tmpC);
memcpy(ctu.m_trCoeff[2] + tmpC2, m_trCoeff[2], sizeof(coeff_t) * tmpC);
}
/* The reverse of copyToPic, called only by encodeResidue */
void CUData::copyFromPic(const CUData& ctu, const CUGeom& cuGeom)
{
m_encData = ctu.m_encData;
m_slice = ctu.m_slice;
m_cuAddr = ctu.m_cuAddr;
m_cuPelX = ctu.m_cuPelX + g_zscanToPelX[cuGeom.absPartIdx];
m_cuPelY = ctu.m_cuPelY + g_zscanToPelY[cuGeom.absPartIdx];
m_absIdxInCTU = cuGeom.absPartIdx;
m_numPartitions = cuGeom.numPartitions;
/* copy out all prediction info for this part */
m_partCopy((uint8_t*)m_qp, (uint8_t*)ctu.m_qp + m_absIdxInCTU);
m_partCopy(m_log2CUSize, ctu.m_log2CUSize + m_absIdxInCTU);
m_partCopy(m_lumaIntraDir, ctu.m_lumaIntraDir + m_absIdxInCTU);
m_partCopy(m_tqBypass, ctu.m_tqBypass + m_absIdxInCTU);
m_partCopy((uint8_t*)m_refIdx[0], (uint8_t*)ctu.m_refIdx[0] + m_absIdxInCTU);
m_partCopy((uint8_t*)m_refIdx[1], (uint8_t*)ctu.m_refIdx[1] + m_absIdxInCTU);
m_partCopy(m_cuDepth, ctu.m_cuDepth + m_absIdxInCTU);
m_partSet(m_predMode, ctu.m_predMode[m_absIdxInCTU] & (MODE_INTRA | MODE_INTER)); /* clear skip flag */
m_partCopy(m_partSize, ctu.m_partSize + m_absIdxInCTU);
m_partCopy(m_mergeFlag, ctu.m_mergeFlag + m_absIdxInCTU);
m_partCopy(m_interDir, ctu.m_interDir + m_absIdxInCTU);
m_partCopy(m_mvpIdx[0], ctu.m_mvpIdx[0] + m_absIdxInCTU);
m_partCopy(m_mvpIdx[1], ctu.m_mvpIdx[1] + m_absIdxInCTU);
m_partCopy(m_chromaIntraDir, ctu.m_chromaIntraDir + m_absIdxInCTU);
memcpy(m_mv[0], ctu.m_mv[0] + m_absIdxInCTU, m_numPartitions * sizeof(MV));
memcpy(m_mv[1], ctu.m_mv[1] + m_absIdxInCTU, m_numPartitions * sizeof(MV));
memcpy(m_mvd[0], ctu.m_mvd[0] + m_absIdxInCTU, m_numPartitions * sizeof(MV));
memcpy(m_mvd[1], ctu.m_mvd[1] + m_absIdxInCTU, m_numPartitions * sizeof(MV));
/* clear residual coding flags */
m_partSet(m_tuDepth, 0);
m_partSet(m_transformSkip[0], 0);
m_partSet(m_transformSkip[1], 0);
m_partSet(m_transformSkip[2], 0);
m_partSet(m_cbf[0], 0);
m_partSet(m_cbf[1], 0);
m_partSet(m_cbf[2], 0);
}
/* Only called by encodeResidue, these fields can be modified during inter/intra coding */
void CUData::updatePic(uint32_t depth) const
{
CUData& ctu = *m_encData->getPicCTU(m_cuAddr);
m_partCopy((uint8_t*)ctu.m_qp + m_absIdxInCTU, (uint8_t*)m_qp);
m_partCopy(ctu.m_transformSkip[0] + m_absIdxInCTU, m_transformSkip[0]);
m_partCopy(ctu.m_transformSkip[1] + m_absIdxInCTU, m_transformSkip[1]);
m_partCopy(ctu.m_transformSkip[2] + m_absIdxInCTU, m_transformSkip[2]);
m_partCopy(ctu.m_predMode + m_absIdxInCTU, m_predMode);
m_partCopy(ctu.m_tuDepth + m_absIdxInCTU, m_tuDepth);
m_partCopy(ctu.m_cbf[0] + m_absIdxInCTU, m_cbf[0]);
m_partCopy(ctu.m_cbf[1] + m_absIdxInCTU, m_cbf[1]);
m_partCopy(ctu.m_cbf[2] + m_absIdxInCTU, m_cbf[2]);
m_partCopy(ctu.m_chromaIntraDir + m_absIdxInCTU, m_chromaIntraDir);
uint32_t tmpY = 1 << ((g_maxLog2CUSize - depth) * 2);
uint32_t tmpY2 = m_absIdxInCTU << (LOG2_UNIT_SIZE * 2);
memcpy(ctu.m_trCoeff[0] + tmpY2, m_trCoeff[0], sizeof(coeff_t) * tmpY);
tmpY >>= m_hChromaShift + m_vChromaShift;
tmpY2 >>= m_hChromaShift + m_vChromaShift;
memcpy(ctu.m_trCoeff[1] + tmpY2, m_trCoeff[1], sizeof(coeff_t) * tmpY);
memcpy(ctu.m_trCoeff[2] + tmpY2, m_trCoeff[2], sizeof(coeff_t) * tmpY);
}
/** 函数功能 ： 获取参考点的CTU或者CU位置，获取参考点在获取CTU或者CU内部的zigzag标号位置
* \参数 lPartUnitIdx ： 用于存储参考像素点在返回CU/CTU 内部的zigzag标号 1.参考像素点不在当前CTU：返回参考像素点所在CTU的zigzag标号2.参考像素点在当前CTU不在当前CU：返回参考像素点在当前CTU的zigzag标号 3.参考像素点在当前CU：回参考像素点在当前CU的zigzag标号
* \参数 curPartUnitIdx ： 当前PU左边行某一4x4块在CTU中的zigzag标号（PU内部的左边）
* 返回值 ： 1.参考像素点不在当前CTU：返回参考的所在的CTU 2.参考像素点在当前CTU不在当前CU：返回当前CTU 3.参考像素点在当前CU：返回当前CU
**/
const CUData* CUData::getPULeft(uint32_t& lPartUnitIdx, uint32_t curPartUnitIdx) const
{
uint32_t absPartIdx = g_zscanToRaster[curPartUnitIdx];//获取当前光栅扫描位置
if (!isZeroCol(absPartIdx, s_numPartInCUSize))//如果不是CTU列中的第一列 说明参考点在CTU内部
{
uint32_t absZorderCUIdx = g_zscanToRaster[m_absIdxInCTU];//将当前CU在CTU位置的zigzag扫描改为光栅扫描
lPartUnitIdx = g_rasterToZscan[absPartIdx - 1];//获取当前PU左边参考像素点的zigzag扫描位置
if (isEqualCol(absPartIdx, absZorderCUIdx, s_numPartInCUSize))//当前PU与CU左上角像素点在同一列 说明参考点不在当前CU
return m_encData->getPicCTU(m_cuAddr);//返回当前CTU
else
{
lPartUnitIdx -= m_absIdxInCTU;//获取在CU内部的参考像素点的zigzag标号
return this;//返回当前CU
}
}
lPartUnitIdx = g_rasterToZscan[absPartIdx + s_numPartInCUSize - 1];//获取在左边CTU的内部zigzag编号
return m_cuLeft;//返回左边的CTU
}
/** 函数功能 ： 获取参考点的CTU或者CU位置，获取参考点在获取CTU或者CU内部的zigzag标号位置
* \参数 alPartUnitIdx ： 用于存储参考像素点在返回CU/CTU 内部的zigzag标号 1.参考像素点不在当前CTU：返回参考像素点所在CTU的zigzag标号2.参考像素点在当前CTU不在当前CU：返回参考像素点在当前CTU的zigzag标号 3.参考像素点在当前CU：回参考像素点在当前CU的zigzag标号
* \参数 curPartUnitIdx ： 当前PU上边行某一4x4块在CTU中的zigzag标号
* 返回值 ： 1.参考像素点不在当前CTU：返回参考的所在的CTU 2.参考像素点在当前CTU不在当前CU：返回当前CTU 3.参考像素点在当前CU：返回当前CU
**/
const CUData* CUData::getPUAbove(uint32_t& aPartUnitIdx, uint32_t curPartUnitIdx) const
{
uint32_t absPartIdx = g_zscanToRaster[curPartUnitIdx];//获取当前光栅扫描位置
if (!isZeroRow(absPartIdx, s_numPartInCUSize))//如果当前不是CTU第一行 （参考像素点在CTU内部）
{
uint32_t absZorderCUIdx = g_zscanToRaster[m_absIdxInCTU];//将当前CU在CTU位置的zigzag扫描改为光栅扫描
aPartUnitIdx = g_rasterToZscan[absPartIdx - s_numPartInCUSize];//获取当前PU上边参考像素点的zigzag扫描位置
if (isEqualRow(absPartIdx, absZorderCUIdx, s_numPartInCUSize))//如果PU的第一行跟CU块的第一行重合 （说明参考像素点在另一个CU）
return m_encData->getPicCTU(m_cuAddr);//返回当前CTU
else
aPartUnitIdx -= m_absIdxInCTU;//获取在CU内部的参考像素点的zigzag标号
return this;//返回当前CU
}
//当前是CTU的第一行 上边行参考像素点在另一个CTU上
aPartUnitIdx = g_rasterToZscan[absPartIdx + NUM_4x4_PARTITIONS - s_numPartInCUSize];//获取在上边CTU的内部zigzag编号
return m_cuAbove;//返回上边的CTU
}
/** 函数功能 ： 获取参考点的CTU或者CU位置，获取参考点在获取CTU或者CU内部的zigzag标号位置
* \参数 alPartUnitIdx ： 用于存储参考像素点在返回CU/CTU 内部的zigzag标号 1.参考像素点不在当前CTU：返回参考像素点所在CTU的zigzag标号2.参考像素点在当前CTU不在当前CU：返回参考像素点在当前CTU的zigzag标号 3.参考像素点在当前CU：回参考像素点在当前CU的zigzag标号
* \参数 curPartUnitIdx ： 当前PU左上角像素点在CTU中的zigzag标号
* 返回值 ： 1.参考像素点不在当前CTU：返回参考的所在的CTU2.参考像素点在当前CTU不在当前CU：返回当前CTU 3.参考像素点在当前CU：返回当前CU
**/
const CUData* CUData::getPUAboveLeft(uint32_t& alPartUnitIdx, uint32_t curPartUnitIdx) const
{
uint32_t absPartIdx = g_zscanToRaster[curPartUnitIdx];//获取当前光栅扫描位置
//以下if 保证左上角参考像素点在本CTU中
if (!isZeroCol(absPartIdx, s_numPartInCUSize))//如果当前不是CTU第一列
{
if (!isZeroRow(absPartIdx, s_numPartInCUSize))//如果当前不是CTU第一行
{
uint32_t absZorderCUIdx = g_zscanToRaster[m_absIdxInCTU];//将当前CU在CTU位置的zigzag扫描改为光栅扫描
alPartUnitIdx = g_rasterToZscan[absPartIdx - s_numPartInCUSize - 1];//获取当前PU左上角参考像素点的zigzag扫描位置
/*以intra NxN为例
----------------
| PU0 | PU1|
----------------
| PU2 | PU3|
----------------
isEqualRowOrCol指的是PU0 和 PU1 PU2 因为其左上角参考像素点在另一个CU中
**/
if (isEqualRowOrCol(absPartIdx, absZorderCUIdx, s_numPartInCUSize))//如果当前PU跟CU是同一行或者同一列
return m_encData->getPicCTU(m_cuAddr);//返回当前所在的CTU
else
{
alPartUnitIdx -= m_absIdxInCTU;//获取在CU内部的参考像素点的zigzag标号
return this;//返回当前CU
}
}
//不在CTU第一列 但在CTU第一行
alPartUnitIdx = g_rasterToZscan[absPartIdx + NUM_4x4_PARTITIONS - s_numPartInCUSize - 1];//获取参考像素点在上边CTU的内部zigzag标号
return m_cuAbove;//返回上边的CTU
}
if (!isZeroRow(absPartIdx, s_numPartInCUSize))//是CTU第一列 不是CTU第一行
{
alPartUnitIdx = g_rasterToZscan[absPartIdx - 1];//获取参考像素点在左边CTU的内部zigzag标号
return m_cuLeft;//返回左边的CTU
}
//当前PU左上角像素点与CTU左上角像素点是同一个位置
alPartUnitIdx = g_rasterToZscan[NUM_4x4_PARTITIONS - 1];//获取参考像素点在左上边CTU的内部zigzag标号
return m_cuAboveLeft;//返回左上边的CTU
}
const CUData* CUData::getPUAboveRight(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx) const
{
if ((m_encData->getPicCTU(m_cuAddr)->m_cuPelX + g_zscanToPelX[curPartUnitIdx] + UNIT_SIZE) >= m_slice->m_sps->picWidthInLumaSamples)
return NULL;
uint32_t absPartIdxRT = g_zscanToRaster[curPartUnitIdx];
if (lessThanCol(absPartIdxRT, s_numPartInCUSize - 1, s_numPartInCUSize))
{
if (!isZeroRow(absPartIdxRT, s_numPartInCUSize))
{
if (curPartUnitIdx > g_rasterToZscan[absPartIdxRT - s_numPartInCUSize + 1])
{
uint32_t absZorderCUIdx = g_zscanToRaster[m_absIdxInCTU] + (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1;
arPartUnitIdx = g_rasterToZscan[absPartIdxRT - s_numPartInCUSize + 1];
if (isEqualRowOrCol(absPartIdxRT, absZorderCUIdx, s_numPartInCUSize))
return m_encData->getPicCTU(m_cuAddr);
else
{
arPartUnitIdx -= m_absIdxInCTU;
return this;
}
}
return NULL;
}
arPartUnitIdx = g_rasterToZscan[absPartIdxRT + NUM_4x4_PARTITIONS - s_numPartInCUSize + 1];
return m_cuAbove;
}
if (!isZeroRow(absPartIdxRT, s_numPartInCUSize))
return NULL;
arPartUnitIdx = g_rasterToZscan[NUM_4x4_PARTITIONS - s_numPartInCUSize];
return m_cuAboveRight;
}
const CUData* CUData::getPUBelowLeft(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx) const
{
if ((m_encData->getPicCTU(m_cuAddr)->m_cuPelY + g_zscanToPelY[curPartUnitIdx] + UNIT_SIZE) >= m_slice->m_sps->picHeightInLumaSamples)
return NULL;
uint32_t absPartIdxLB = g_zscanToRaster[curPartUnitIdx];
if (lessThanRow(absPartIdxLB, s_numPartInCUSize - 1, s_numPartInCUSize))
{
if (!isZeroCol(absPartIdxLB, s_numPartInCUSize))
{
if (curPartUnitIdx > g_rasterToZscan[absPartIdxLB + s_numPartInCUSize - 1])
{
uint32_t absZorderCUIdxLB = g_zscanToRaster[m_absIdxInCTU] + ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1) * s_numPartInCUSize;
blPartUnitIdx = g_rasterToZscan[absPartIdxLB + s_numPartInCUSize - 1];
if (isEqualRowOrCol(absPartIdxLB, absZorderCUIdxLB, s_numPartInCUSize))
return m_encData->getPicCTU(m_cuAddr);
else
{
blPartUnitIdx -= m_absIdxInCTU;
return this;
}
}
return NULL;
}
blPartUnitIdx = g_rasterToZscan[absPartIdxLB + s_numPartInCUSize * 2 - 1];
return m_cuLeft;
}
return NULL;
}
/** 函数功能 ： 获取参考点的CTU或者CU位置，获取参考点在获取CTU或者CU内部的zigzag标号位置
* \参数 blPartUnitIdx ： 用于存储参考像素点在返回CU/CTU 内部的zigzag标号 1.参考像素点不在当前CTU：返回参考像素点所在CTU的zigzag标号2.参考像素点在当前CTU不在当前CU：返回参考像素点在当前CTU的zigzag标号 3.参考像素点在当前CU：回参考像素点在当前CU的zigzag标号
* \参数 curPartUnitIdx ： 当前PU右上角像素点在CTU中的zigzag标号
* \参数 partUnitOffset ： 当前4x4块距离PU左下点位置的偏移值
* 返回值 ： 1.参考像素点不在当前CTU：返回参考的所在的CTU 2.参考像素点在当前CTU不在当前CU：返回当前CTU 3.参考像素点在当前CU：返回当前CU
**/
const CUData* CUData::getPUBelowLeftAdi(uint32_t& blPartUnitIdx, uint32_t curPartUnitIdx, uint32_t partUnitOffset) const
{
if ((m_encData->getPicCTU(m_cuAddr)->m_cuPelY + g_zscanToPelY[curPartUnitIdx] + (partUnitOffset << LOG2_UNIT_SIZE)) >= m_slice->m_sps->picHeightInLumaSamples)//判断是否越过图像边界
return NULL;
uint32_t absPartIdxLB = g_zscanToRaster[curPartUnitIdx];//获取当前光栅扫描位置
if (lessThanRow(absPartIdxLB, s_numPartInCUSize - partUnitOffset, s_numPartInCUSize))//如果被参考块没有超过当前CTU行的边界 （在同一个CTU中）
{
if (!isZeroCol(absPartIdxLB, s_numPartInCUSize))//如果当前不是CTU第一列
{
if (curPartUnitIdx > g_rasterToZscan[absPartIdxLB + partUnitOffset * s_numPartInCUSize - 1])//必须保证左下块已经编码完毕 即当前的zigzag号要大于参考块的zigzag号（因为当前在同一个CTU下）
{
uint32_t absZorderCUIdxLB = g_zscanToRaster[m_absIdxInCTU] + ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1) * s_numPartInCUSize;//将当前CU左下角在CTU位置的zigzag扫描改为光栅扫描
blPartUnitIdx = g_rasterToZscan[absPartIdxLB + partUnitOffset * s_numPartInCUSize - 1];//获取左下位置
if (isEqualRowOrCol(absPartIdxLB, absZorderCUIdxLB, s_numPartInCUSize))//如果当前PU左下角与CU处于同一行 或者同一列
return m_encData->getPicCTU(m_cuAddr);//返回当前CTU
else
{
//同一CU下不可能获取左下可用块
blPartUnitIdx -= m_absIdxInCTU;//获取在CU内部的参考像素点的zigzag标号
return this;//返回当前CU
}
}
return NULL;//左下不可用 返回null
}
blPartUnitIdx = g_rasterToZscan[absPartIdxLB + (1 + partUnitOffset) * s_numPartInCUSize - 1];//获取参考像素点在左边CTU的内部zigzag标号
return m_cuLeft;//返回左边CTU
}
return NULL;//返回null
}
/** 函数功能 ： 获取参考点的CTU或者CU位置，获取参考点在获取CTU或者CU内部的zigzag标号位置
* \参数 alPartUnitIdx ： 用于存储参考像素点在返回CU/CTU 内部的zigzag标号 1.参考像素点不在当前CTU：返回参考像素点所在CTU的zigzag标号2.参考像素点在当前CTU不在当前CU：返回参考像素点在当前CTU的zigzag标号 3.参考像素点在当前CU：回参考像素点在当前CU的zigzag标号
* \参数 curPartUnitIdx ： 当前PU右上角像素点在CTU中的zigzag标号
* \参数 partUnitOffset ： 当前4x4块距离PU右上点位置的偏移值
* 返回值 ： 1.参考像素点不在当前CTU：返回参考的所在的CTU 2.参考像素点在当前CTU不在当前CU：返回当前CTU 3.参考像素点在当前CU：返回当前CU
**/
const CUData* CUData::getPUAboveRightAdi(uint32_t& arPartUnitIdx, uint32_t curPartUnitIdx, uint32_t partUnitOffset) const
{
if ((m_encData->getPicCTU(m_cuAddr)->m_cuPelX + g_zscanToPelX[curPartUnitIdx] + (partUnitOffset << LOG2_UNIT_SIZE)) >= m_slice->m_sps->picWidthInLumaSamples)//判断是否越过图像边界
return NULL;
uint32_t absPartIdxRT = g_zscanToRaster[curPartUnitIdx];//获取当前光栅扫描位置
if (lessThanCol(absPartIdxRT, s_numPartInCUSize - partUnitOffset, s_numPartInCUSize))//如果被参考块没有超过当前CTU列的边界 （在同一个CTU中）
{
if (!isZeroRow(absPartIdxRT, s_numPartInCUSize))//如果当前不是CTU第一行
{
if (curPartUnitIdx > g_rasterToZscan[absPartIdxRT - s_numPartInCUSize + partUnitOffset])//必须保证右上块已经编码完毕 即当前的zigzag号要大于参考块的zigzag号（因为当前在同一个CTU下）
{
uint32_t absZorderCUIdx = g_zscanToRaster[m_absIdxInCTU] + (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1;//将当前CU左上角在CTU位置的zigzag扫描改为光栅扫描
arPartUnitIdx = g_rasterToZscan[absPartIdxRT - s_numPartInCUSize + partUnitOffset];//获取右上位置
/*以intra NxN为例
----------------
| PU0 | PU1|
----------------
| PU2 | PU3|
----------------
isEqualRowOrCol指的是PU0 和 PU1 PU3 因为其右上角参考像素点在另一个CU中
注意：这里指的是PU右上角像素点
**/
if (isEqualRowOrCol(absPartIdxRT, absZorderCUIdx, s_numPartInCUSize))//如果当前PU右上角与CU处于同一行 或者同一列
return m_encData->getPicCTU(m_cuAddr);//返回当前CTU
else
{
arPartUnitIdx -= m_absIdxInCTU;//获取在CU内部的参考像素点的zigzag标号
return this;//返回当前CU
}
}
return NULL;//右上不可用 返回null
}
//当前PU右上角像素点是CTU的第一行
arPartUnitIdx = g_rasterToZscan[absPartIdxRT + NUM_4x4_PARTITIONS - s_numPartInCUSize + partUnitOffset];//获取参考像素点在上边CTU的内部zigzag标号
return m_cuAbove;//返回上边CTU （因为当前右上角参考像素行在上边CTU中）
}
if (!isZeroRow(absPartIdxRT, s_numPartInCUSize))//当前PU不在CTU第一行 说明右上参考像素点在右边CTU（尚未编码）
return NULL;//返回null
arPartUnitIdx = g_rasterToZscan[NUM_4x4_PARTITIONS - s_numPartInCUSize + partUnitOffset - 1];//获取参考像素点在右上边CTU的内部zigzag标号
return m_cuAboveRight;//返回右上CTU
}
/* Get left QpMinCu */
const CUData* CUData::getQpMinCuLeft(uint32_t& lPartUnitIdx, uint32_t curAbsIdxInCTU) const
{
uint32_t absZorderQpMinCUIdx = curAbsIdxInCTU & (0xFF << (g_unitSizeDepth - m_slice->m_pps->maxCuDQPDepth) * 2);
uint32_t absRorderQpMinCUIdx = g_zscanToRaster[absZorderQpMinCUIdx];
// check for left CTU boundary
if (isZeroCol(absRorderQpMinCUIdx, s_numPartInCUSize))
return NULL;
// get index of left-CU relative to top-left corner of current quantization group
lPartUnitIdx = g_rasterToZscan[absRorderQpMinCUIdx - 1];
// return pointer to current CTU
return m_encData->getPicCTU(m_cuAddr);
}
/* Get above QpMinCu */
const CUData* CUData::getQpMinCuAbove(uint32_t& aPartUnitIdx, uint32_t curAbsIdxInCTU) const
{
uint32_t absZorderQpMinCUIdx = curAbsIdxInCTU & (0xFF << (g_unitSizeDepth - m_slice->m_pps->maxCuDQPDepth) * 2);
uint32_t absRorderQpMinCUIdx = g_zscanToRaster[absZorderQpMinCUIdx];
// check for top CTU boundary
if (isZeroRow(absRorderQpMinCUIdx, s_numPartInCUSize))
return NULL;
// get index of top-CU relative to top-left corner of current quantization group
aPartUnitIdx = g_rasterToZscan[absRorderQpMinCUIdx - s_numPartInCUSize];
// return pointer to current CTU
return m_encData->getPicCTU(m_cuAddr);
}
/* Get reference QP from left QpMinCu or latest coded QP */
int8_t CUData::getRefQP(uint32_t curAbsIdxInCTU) const
{
uint32_t lPartIdx = 0, aPartIdx = 0;
const CUData* cULeft = getQpMinCuLeft(lPartIdx, m_absIdxInCTU + curAbsIdxInCTU);
const CUData* cUAbove = getQpMinCuAbove(aPartIdx, m_absIdxInCTU + curAbsIdxInCTU);
return ((cULeft ? cULeft->m_qp[lPartIdx] : getLastCodedQP(curAbsIdxInCTU)) + (cUAbove ? cUAbove->m_qp[aPartIdx] : getLastCodedQP(curAbsIdxInCTU)) + 1) >> 1;
}
int CUData::getLastValidPartIdx(int absPartIdx) const
{
int lastValidPartIdx = absPartIdx - 1;
while (lastValidPartIdx >= 0 && m_predMode[lastValidPartIdx] == MODE_NONE)
{
uint32_t depth = m_cuDepth[lastValidPartIdx];
lastValidPartIdx -= m_numPartitions >> (depth << 1);
}
return lastValidPartIdx;
}
int8_t CUData::getLastCodedQP(uint32_t absPartIdx) const
{
uint32_t quPartIdxMask = 0xFF << (g_unitSizeDepth - m_slice->m_pps->maxCuDQPDepth) * 2;
int lastValidPartIdx = getLastValidPartIdx(absPartIdx & quPartIdxMask);
if (lastValidPartIdx >= 0)
return m_qp[lastValidPartIdx];
else
{
if (m_absIdxInCTU)
return m_encData->getPicCTU(m_cuAddr)->getLastCodedQP(m_absIdxInCTU);
else if (m_cuAddr > 0 && !(m_slice->m_pps->bEntropyCodingSyncEnabled && !(m_cuAddr % m_slice->m_sps->numCuInWidth)))
return m_encData->getPicCTU(m_cuAddr - 1)->getLastCodedQP(NUM_4x4_PARTITIONS);
else
return (int8_t)m_slice->m_sliceQp;
}
}
/* Get allowed chroma intra modes */
void CUData::getAllowedChromaDir(uint32_t absPartIdx, uint32_t* modeList) const
{
modeList[0] = PLANAR_IDX;
modeList[1] = VER_IDX;
modeList[2] = HOR_IDX;
modeList[3] = DC_IDX;
modeList[4] = DM_CHROMA_IDX;
uint32_t lumaMode = m_lumaIntraDir[absPartIdx];
for (int i = 0; i < NUM_CHROMA_MODE - 1; i++)
{
if (lumaMode == modeList[i])
{
modeList[i] = 34; // VER+8 mode
break;
}
}
}
/* Get most probable intra modes */
/** 函数功能 ： 获取3个最有可能的帧内预测模式 并A（左边）与B（上边）方向相同（或者都不是intra块）则返回1 不同返回2
/* 调用范围 ： 只在codeIntraDirLumaAng和getIntraRemModeBits函数中被调用
* \参数 absPartIdx ： PU在CU的zigzag标号
* \参数 intraDirPred ： 存储3个最有可能的帧内预测模式
* 返回值 ： A（左边）与B（上边）方向相同（或者都不是intra块）则返回1 不同返回2**/
int CUData::getIntraDirLumaPredictor(uint32_t absPartIdx, uint32_t* intraDirPred) const
{
const CUData* tempCU;//用于临时存储CU
uint32_t tempPartIdx;//用于临时存储相邻块的zigzag编号
uint32_t leftIntraDir, aboveIntraDir;//分别存储 左边 和上边的预测方向
/* B：aboveIntraDir A：leftIntraDir
B
-------------
A | |
| 当前PU |
-------------
**/
// Get intra direction of left PU
tempCU = getPULeft(tempPartIdx, m_absIdxInCTU + absPartIdx);//获取A所在的CU：tempCU tempPartIdx：A在tempCU的zigzag编号
leftIntraDir = (tempCU && tempCU->isIntra(tempPartIdx)) ? tempCU->m_lumaIntraDir[tempPartIdx] : DC_IDX;//获取A的方向 无则置为DC
// Get intra direction of above PU
tempCU = g_zscanToPelY[m_absIdxInCTU + absPartIdx] > 0 ? getPUAbove(tempPartIdx, m_absIdxInCTU + absPartIdx) : NULL;//获取B所在的CU：tempCU tempPartIdx：B在tempCU的zigzag编号
aboveIntraDir = (tempCU && tempCU->isIntra(tempPartIdx)) ? tempCU->m_lumaIntraDir[tempPartIdx] : DC_IDX;//获取B的方向 无则置为DC
if (leftIntraDir == aboveIntraDir)//如果A与B的方向相同
{
if (leftIntraDir >= 2) //如果A与B都为角度模式 angular modes
{
/*角度模式：
(左33) 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 (右3)
**/
intraDirPred[0] = leftIntraDir; //0：获取A的方向
intraDirPred[1] = ((leftIntraDir - 2 + 31) & 31) + 2;//1:获取A相邻的左边方向
intraDirPred[2] = ((leftIntraDir - 2 + 1) & 31) + 2;//2：获取A相邻的右边方向
}
else //如果A与B都不为角度模式non-angular
{
intraDirPred[0] = PLANAR_IDX;//0：为 PLANar模式
intraDirPred[1] = DC_IDX;//1:DC模式
intraDirPred[2] = VER_IDX;//2:垂直模式
}
return 1;//A（左边）与B（上边）方向相同（或者都不是intra块）则返回1
}
else//如果A与B的方向不同
{
intraDirPred[0] = leftIntraDir;//0：A 左边预测方向
intraDirPred[1] = aboveIntraDir;//1：B 上边预测方向
if (leftIntraDir && aboveIntraDir) //A与B任意一个都不为Planar模式 both modes are non-planar
intraDirPred[2] = PLANAR_IDX;//2:置为planar模式
else
intraDirPred[2] = (leftIntraDir + aboveIntraDir) < 2 ? VER_IDX : DC_IDX;//A与B任意一个都不为DC或者PLANAR 则2：DC 否则 2：垂直模式 注意：因为此处已经排除A=B=DC模式 所有 小于2即可
return 2;//A（左边）与B（上边）方向不同 则返回1
}
}
uint32_t CUData::getCtxSplitFlag(uint32_t absPartIdx, uint32_t depth) const
{
const CUData* tempCU;
uint32_t tempPartIdx;
uint32_t ctx;
// Get left split flag
tempCU = getPULeft(tempPartIdx, m_absIdxInCTU + absPartIdx);
ctx = (tempCU) ? ((tempCU->m_cuDepth[tempPartIdx] > depth) ? 1 : 0) : 0;
// Get above split flag
tempCU = getPUAbove(tempPartIdx, m_absIdxInCTU + absPartIdx);
ctx += (tempCU) ? ((tempCU->m_cuDepth[tempPartIdx] > depth) ? 1 : 0) : 0;
return ctx;
}
/** 函数功能 ： 分别获取intraCU下最小TU，最大TU
* \参数 tuDepthRange[2]： 分别存储最小TU，最大TU
* \参数 absPartIdx ： 在Entropy::encodeCU为当前CU的左上角标号，其它位置为0
* 返回值 ： null
**/
void CUData::getIntraTUQtDepthRange(uint32_t tuDepthRange[2], uint32_t absPartIdx) const
{
/*
样例：m_slice->m_sps->quadtreeTULog2MinSize =2 m_slice->m_sps->quadtreeTULog2MaxSize = 5 m_slice->m_sps->quadtreeTUMaxDepthIntra =2
cu:6 split:0 tu[0]=5 tu[1]=5
cu:6 split:1 tu[0]=4 tu[1]=5
cu:5 split:0 tu[0]=4 tu[1]=5
cu:5 split:1 tu[0]=3 tu[1]=5
cu:4 split:0 tu[0]=3 tu[1]=5
cu:4 split:1 tu[0]=2 tu[1]=5
cu:3 split:0 tu[0]=2 tu[1]=5
cu:3 split:1 tu[0]=2 tu[1]=5
**/
uint32_t log2CUSize = m_log2CUSize[absPartIdx];//获取当前CU大小 6:64x64 5:32x32 .....
uint32_t splitFlag = m_partSize[absPartIdx] != SIZE_2Nx2N;//是否划分 2Nx2N 为0 其它为1
tuDepthRange[0] = m_slice->m_sps->quadtreeTULog2MinSize; // 最大TU尺寸
tuDepthRange[1] = m_slice->m_sps->quadtreeTULog2MaxSize; // 最小TU尺寸
tuDepthRange[0] = x265_clip3(tuDepthRange[0], tuDepthRange[1], log2CUSize - (m_slice->m_sps->quadtreeTUMaxDepthIntra - 1 + splitFlag));//获取最小TU CU-（TUintra - 1 + split)
}
void CUData::getInterTUQtDepthRange(uint32_t tuDepthRange[2], uint32_t absPartIdx) const
{
uint32_t log2CUSize = m_log2CUSize[absPartIdx];
uint32_t quadtreeTUMaxDepth = m_slice->m_sps->quadtreeTUMaxDepthInter;
uint32_t splitFlag = quadtreeTUMaxDepth == 1 && m_partSize[absPartIdx] != SIZE_2Nx2N;
tuDepthRange[0] = m_slice->m_sps->quadtreeTULog2MinSize;
tuDepthRange[1] = m_slice->m_sps->quadtreeTULog2MaxSize;
tuDepthRange[0] = x265_clip3(tuDepthRange[0], tuDepthRange[1], log2CUSize - (quadtreeTUMaxDepth - 1 + splitFlag));
}
uint32_t CUData::getCtxSkipFlag(uint32_t absPartIdx) const
{
const CUData* tempCU;
uint32_t tempPartIdx;
uint32_t ctx;
// Get BCBP of left PU
tempCU = getPULeft(tempPartIdx, m_absIdxInCTU + absPartIdx);
ctx = tempCU ? tempCU->isSkipped(tempPartIdx) : 0;
// Get BCBP of above PU
tempCU = getPUAbove(tempPartIdx, m_absIdxInCTU + absPartIdx);
ctx += tempCU ? tempCU->isSkipped(tempPartIdx) : 0;
return ctx;
}
bool CUData::setQPSubCUs(int8_t qp, uint32_t absPartIdx, uint32_t depth)
{
uint32_t curPartNumb = NUM_4x4_PARTITIONS >> (depth << 1);
uint32_t curPartNumQ = curPartNumb >> 2;
if (m_cuDepth[absPartIdx] > depth)
{
for (uint32_t subPartIdx = 0; subPartIdx < 4; subPartIdx++)
if (setQPSubCUs(qp, absPartIdx + subPartIdx * curPartNumQ, depth + 1))
return true;
}
else
{
if (getQtRootCbf(absPartIdx))
return true;
else
setQPSubParts(qp, absPartIdx, depth);
}
return false;
}
void CUData::setPUInterDir(uint8_t dir, uint32_t absPartIdx, uint32_t puIdx)
{
uint32_t curPartNumQ = m_numPartitions >> 2;
X265_CHECK(puIdx < 2, "unexpected part unit index\n");
switch (m_partSize[absPartIdx])
{
case SIZE_2Nx2N:
memset(m_interDir + absPartIdx, dir, 4 * curPartNumQ);
break;
case SIZE_2NxN:
memset(m_interDir + absPartIdx, dir, 2 * curPartNumQ);
break;
case SIZE_Nx2N:
memset(m_interDir + absPartIdx, dir, curPartNumQ);
memset(m_interDir + absPartIdx + 2 * curPartNumQ, dir, curPartNumQ);
break;
case SIZE_NxN:
memset(m_interDir + absPartIdx, dir, curPartNumQ);
break;
case SIZE_2NxnU:
if (!puIdx)
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ >> 1));
memset(m_interDir + absPartIdx + curPartNumQ, dir, (curPartNumQ >> 1));
}
else
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ >> 1));
memset(m_interDir + absPartIdx + curPartNumQ, dir, ((curPartNumQ >> 1) + (curPartNumQ << 1)));
}
break;
case SIZE_2NxnD:
if (!puIdx)
{
memset(m_interDir + absPartIdx, dir, ((curPartNumQ << 1) + (curPartNumQ >> 1)));
memset(m_interDir + absPartIdx + (curPartNumQ << 1) + curPartNumQ, dir, (curPartNumQ >> 1));
}
else
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ >> 1));
memset(m_interDir + absPartIdx + curPartNumQ, dir, (curPartNumQ >> 1));
}
break;
case SIZE_nLx2N:
if (!puIdx)
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ >> 1), dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ << 1), dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ << 1) + (curPartNumQ >> 1), dir, (curPartNumQ >> 2));
}
else
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ >> 1), dir, (curPartNumQ + (curPartNumQ >> 2)));
memset(m_interDir + absPartIdx + (curPartNumQ << 1), dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ << 1) + (curPartNumQ >> 1), dir, (curPartNumQ + (curPartNumQ >> 2)));
}
break;
case SIZE_nRx2N:
if (!puIdx)
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ + (curPartNumQ >> 2)));
memset(m_interDir + absPartIdx + curPartNumQ + (curPartNumQ >> 1), dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ << 1), dir, (curPartNumQ + (curPartNumQ >> 2)));
memset(m_interDir + absPartIdx + (curPartNumQ << 1) + curPartNumQ + (curPartNumQ >> 1), dir, (curPartNumQ >> 2));
}
else
{
memset(m_interDir + absPartIdx, dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ >> 1), dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ << 1), dir, (curPartNumQ >> 2));
memset(m_interDir + absPartIdx + (curPartNumQ << 1) + (curPartNumQ >> 1), dir, (curPartNumQ >> 2));
}
break;
default:
X265_CHECK(0, "unexpected part type\n");
break;
}
}
template<typename T>
void CUData::setAllPU(T* p, const T& val, int absPartIdx, int puIdx)
{
int i;
p += absPartIdx;
int numElements = m_numPartitions;
switch (m_partSize[absPartIdx])
{
case SIZE_2Nx2N:
for (i = 0; i < numElements; i++)
p[i] = val;
break;
case SIZE_2NxN:
numElements >>= 1;
for (i = 0; i < numElements; i++)
p[i] = val;
break;
case SIZE_Nx2N:
numElements >>= 2;
for (i = 0; i < numElements; i++)
{
p[i] = val;
p[i + 2 * numElements] = val;
}
break;
case SIZE_2NxnU:
{
int curPartNumQ = numElements >> 2;
if (!puIdx)
{
T *pT = p;
T *pT2 = p + curPartNumQ;
for (i = 0; i < (curPartNumQ >> 1); i++)
{
pT[i] = val;
pT2[i] = val;
}
}
else
{
T *pT = p;
for (i = 0; i < (curPartNumQ >> 1); i++)
pT[i] = val;
pT = p + curPartNumQ;
for (i = 0; i < ((curPartNumQ >> 1) + (curPartNumQ << 1)); i++)
pT[i] = val;
}
break;
}
case SIZE_2NxnD:
{
int curPartNumQ = numElements >> 2;
if (!puIdx)
{
T *pT = p;
for (i = 0; i < ((curPartNumQ >> 1) + (curPartNumQ << 1)); i++)
pT[i] = val;
pT = p + (numElements - curPartNumQ);
for (i = 0; i < (curPartNumQ >> 1); i++)
pT[i] = val;
}
else
{
T *pT = p;
T *pT2 = p + curPartNumQ;
for (i = 0; i < (curPartNumQ >> 1); i++)
{
pT[i] = val;
pT2[i] = val;
}
}
break;
}
case SIZE_nLx2N:
{
int curPartNumQ = numElements >> 2;
if (!puIdx)
{
T *pT = p;
T *pT2 = p + (curPartNumQ << 1);
T *pT3 = p + (curPartNumQ >> 1);
T *pT4 = p + (curPartNumQ << 1) + (curPartNumQ >> 1);
for (i = 0; i < (curPartNumQ >> 2); i++)
{
pT[i] = val;
pT2[i] = val;
pT3[i] = val;
pT4[i] = val;
}
}
else
{
T *pT = p;
T *pT2 = p + (curPartNumQ << 1);
for (i = 0; i < (curPartNumQ >> 2); i++)
{
pT[i] = val;
pT2[i] = val;
}
pT = p + (curPartNumQ >> 1);
pT2 = p + (curPartNumQ << 1) + (curPartNumQ >> 1);
for (i = 0; i < ((curPartNumQ >> 2) + curPartNumQ); i++)
{
pT[i] = val;
pT2[i] = val;
}
}
break;
}
case SIZE_nRx2N:
{
int curPartNumQ = numElements >> 2;
if (!puIdx)
{
T *pT = p;
T *pT2 = p + (curPartNumQ << 1);
for (i = 0; i < ((curPartNumQ >> 2) + curPartNumQ); i++)
{
pT[i] = val;
pT2[i] = val;
}
pT = p + curPartNumQ + (curPartNumQ >> 1);
pT2 = p + numElements - curPartNumQ + (curPartNumQ >> 1);
for (i = 0; i < (curPartNumQ >> 2); i++)
{
pT[i] = val;
pT2[i] = val;
}
}
else
{
T *pT = p;
T *pT2 = p + (curPartNumQ >> 1);
T *pT3 = p + (curPartNumQ << 1);
T *pT4 = p + (curPartNumQ << 1) + (curPartNumQ >> 1);
for (i = 0; i < (curPartNumQ >> 2); i++)
{
pT[i] = val;
pT2[i] = val;
pT3[i] = val;
pT4[i] = val;
}
}
break;
}
case SIZE_NxN:
default:
X265_CHECK(0, "unknown partition type\n");
break;
}
}
void CUData::setPUMv(int list, const MV& mv, int absPartIdx, int puIdx)
{
setAllPU(m_mv[list], mv, absPartIdx, puIdx);
}
void CUData::setPURefIdx(int list, int8_t refIdx, int absPartIdx, int puIdx)
{
setAllPU(m_refIdx[list], refIdx, absPartIdx, puIdx);
}
void CUData::getPartIndexAndSize(uint32_t partIdx, uint32_t& outPartAddr, int& outWidth, int& outHeight) const
{
int cuSize = 1 << m_log2CUSize[0];
int partType = m_partSize[0];
int tmp = partTable[partType][partIdx][0];
outWidth = ((tmp >> 4) * cuSize) >> 2;
outHeight = ((tmp & 0xF) * cuSize) >> 2;
outPartAddr = (partAddrTable[partType][partIdx] * m_numPartitions) >> 4;
}
void CUData::getMvField(const CUData* cu, uint32_t absPartIdx, int picList, MVField& outMvField) const
{
if (cu)
{
outMvField.mv = cu->m_mv[picList][absPartIdx];
outMvField.refIdx = cu->m_refIdx[picList][absPartIdx];
}
else
{
// OUT OF BOUNDARY
outMvField.mv = 0;
outMvField.refIdx = REF_NOT_VALID;
}
}
void CUData::deriveLeftRightTopIdx(uint32_t partIdx, uint32_t& partIdxLT, uint32_t& partIdxRT) const
{
partIdxLT = m_absIdxInCTU;
partIdxRT = g_rasterToZscan[g_zscanToRaster[partIdxLT] + (1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1];
switch (m_partSize[0])
{
case SIZE_2Nx2N: break;
case SIZE_2NxN:
partIdxLT += (partIdx == 0) ? 0 : m_numPartitions >> 1;
partIdxRT += (partIdx == 0) ? 0 : m_numPartitions >> 1;
break;
case SIZE_Nx2N:
partIdxLT += (partIdx == 0) ? 0 : m_numPartitions >> 2;
partIdxRT -= (partIdx == 1) ? 0 : m_numPartitions >> 2;
break;
case SIZE_NxN:
partIdxLT += (m_numPartitions >> 2) * partIdx;
partIdxRT += (m_numPartitions >> 2) * (partIdx - 1);
break;
case SIZE_2NxnU:
partIdxLT += (partIdx == 0) ? 0 : m_numPartitions >> 3;
partIdxRT += (partIdx == 0) ? 0 : m_numPartitions >> 3;
break;
case SIZE_2NxnD:
partIdxLT += (partIdx == 0) ? 0 : (m_numPartitions >> 1) + (m_numPartitions >> 3);
partIdxRT += (partIdx == 0) ? 0 : (m_numPartitions >> 1) + (m_numPartitions >> 3);
break;
case SIZE_nLx2N:
partIdxLT += (partIdx == 0) ? 0 : m_numPartitions >> 4;
partIdxRT -= (partIdx == 1) ? 0 : (m_numPartitions >> 2) + (m_numPartitions >> 4);
break;
case SIZE_nRx2N:
partIdxLT += (partIdx == 0) ? 0 : (m_numPartitions >> 2) + (m_numPartitions >> 4);
partIdxRT -= (partIdx == 1) ? 0 : m_numPartitions >> 4;
break;
default:
X265_CHECK(0, "unexpected part index\n");
break;
}
}
uint32_t CUData::deriveLeftBottomIdx(uint32_t puIdx) const
{
uint32_t outPartIdxLB;
outPartIdxLB = g_rasterToZscan[g_zscanToRaster[m_absIdxInCTU] + ((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE - 1)) - 1) * s_numPartInCUSize];
switch (m_partSize[0])
{
case SIZE_2Nx2N:
outPartIdxLB += m_numPartitions >> 1;
break;
case SIZE_2NxN:
outPartIdxLB += puIdx ? m_numPartitions >> 1 : 0;
break;
case SIZE_Nx2N:
outPartIdxLB += puIdx ? (m_numPartitions >> 2) * 3 : m_numPartitions >> 1;
break;
case SIZE_NxN:
outPartIdxLB += (m_numPartitions >> 2) * puIdx;
break;
case SIZE_2NxnU:
outPartIdxLB += puIdx ? m_numPartitions >> 1 : -((int)m_numPartitions >> 3);
break;
case SIZE_2NxnD:
outPartIdxLB += puIdx ? m_numPartitions >> 1 : (m_numPartitions >> 2) + (m_numPartitions >> 3);
break;
case SIZE_nLx2N:
outPartIdxLB += puIdx ? (m_numPartitions >> 1) + (m_numPartitions >> 4) : m_numPartitions >> 1;
break;
case SIZE_nRx2N:
outPartIdxLB += puIdx ? (m_numPartitions >> 1) + (m_numPartitions >> 2) + (m_numPartitions >> 4) : m_numPartitions >> 1;
break;
default:
X265_CHECK(0, "unexpected part index\n");
break;
}
return outPartIdxLB;
}
/* Derives the partition index of neighboring bottom right block */
uint32_t CUData::deriveRightBottomIdx(uint32_t puIdx) const
{
uint32_t outPartIdxRB;
outPartIdxRB = g_rasterToZscan[g_zscanToRaster[m_absIdxInCTU] +
((1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE - 1)) - 1) * s_numPartInCUSize +
(1 << (m_log2CUSize[0] - LOG2_UNIT_SIZE)) - 1];
switch (m_partSize[0])
{
case SIZE_2Nx2N:
outPartIdxRB += m_numPartitions >> 1;
break;
case SIZE_2NxN:
outPartIdxRB += puIdx ? m_numPartitions >> 1 : 0;
break;
case SIZE_Nx2N:
outPartIdxRB += puIdx ? m_numPartitions >> 1 : m_numPartitions >> 2;
break;
case SIZE_NxN:
outPartIdxRB += (m_numPartitions >> 2) * (puIdx - 1);
break;
case SIZE_2NxnU:
outPartIdxRB += puIdx ? m_numPartitions >> 1 : -((int)m_numPartitions >> 3);
break;
case SIZE_2NxnD:
outPartIdxRB += puIdx ? m_numPartitions >> 1 : (m_numPartitions >> 2) + (m_numPartitions >> 3);
break;
case SIZE_nLx2N:
outPartIdxRB += puIdx ? m_numPartitions >> 1 : (m_numPartitions >> 3) + (m_numPartitions >> 4);
break;
case SIZE_nRx2N:
outPartIdxRB += puIdx ? m_numPartitions >> 1 : (m_numPartitions >> 2) + (m_numPartitions >> 3) + (m_numPartitions >> 4);
break;
default:
X265_CHECK(0, "unexpected part index\n");
break;
}
return outPartIdxRB;
}
bool CUData::hasEqualMotion(uint32_t absPartIdx, const CUData& candCU, uint32_t candAbsPartIdx) const
{
if (m_interDir[absPartIdx] != candCU.m_interDir[candAbsPartIdx])
return false;
for (uint32_t refListIdx = 0; refListIdx < 2; refListIdx++)
{
if (m_interDir[absPartIdx] & (1 << refListIdx))
{
if (m_mv[refListIdx][absPartIdx] != candCU.m_mv[refListIdx][candAbsPartIdx] ||
m_refIdx[refListIdx][absPartIdx] != candCU.m_refIdx[refListIdx][candAbsPartIdx])
return false;
}
}
return true;
}
/* Construct list of merging candidates, returns count */
uint32_t CUData::getInterMergeCandidates(uint32_t absPartIdx, uint32_t puIdx, MVField(*candMvField)[2], uint8_t* candDir) const
{
uint32_t absPartAddr = m_absIdxInCTU + absPartIdx;
const bool isInterB = m_slice->isInterB();
const uint32_t maxNumMergeCand = m_slice->m_maxNumMergeCand;
for (uint32_t i = 0; i < maxNumMergeCand; ++i)
{
candMvField[i][0].mv = 0;
candMvField[i][1].mv = 0;
candMvField[i][0].refIdx = REF_NOT_VALID;
candMvField[i][1].refIdx = REF_NOT_VALID;
}
/* calculate the location of upper-left corner pixel and size of the current PU */
int xP, yP, nPSW, nPSH;
int cuSize = 1 << m_log2CUSize[0];
int partMode = m_partSize[0];
int tmp = partTable[partMode][puIdx][0];
nPSW = ((tmp >> 4) * cuSize) >> 2;
nPSH = ((tmp & 0xF) * cuSize) >> 2;
tmp = partTable[partMode][puIdx][1];
xP = ((tmp >> 4) * cuSize) >> 2;
yP = ((tmp & 0xF) * cuSize) >> 2;
uint32_t count = 0;
uint32_t partIdxLT, partIdxRT, partIdxLB = deriveLeftBottomIdx(puIdx);
PartSize curPS = (PartSize)m_partSize[absPartIdx];
// left
uint32_t leftPartIdx = 0;
const CUData* cuLeft = getPULeft(leftPartIdx, partIdxLB);
bool isAvailableA1 = cuLeft &&
cuLeft->isDiffMER(xP - 1, yP + nPSH - 1, xP, yP) &&
!(puIdx == 1 && (curPS == SIZE_Nx2N || curPS == SIZE_nLx2N || curPS == SIZE_nRx2N)) &&
cuLeft->isInter(leftPartIdx);
if (isAvailableA1)
{
// get Inter Dir
candDir[count] = cuLeft->m_interDir[leftPartIdx];
// get Mv from Left
cuLeft->getMvField(cuLeft, leftPartIdx, 0, candMvField[count][0]);
if (isInterB)
cuLeft->getMvField(cuLeft, leftPartIdx, 1, candMvField[count][1]);
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
deriveLeftRightTopIdx(puIdx, partIdxLT, partIdxRT);
// above
uint32_t abovePartIdx = 0;
const CUData* cuAbove = getPUAbove(abovePartIdx, partIdxRT);
bool isAvailableB1 = cuAbove &&
cuAbove->isDiffMER(xP + nPSW - 1, yP - 1, xP, yP) &&
!(puIdx == 1 && (curPS == SIZE_2NxN || curPS == SIZE_2NxnU || curPS == SIZE_2NxnD)) &&
cuAbove->isInter(abovePartIdx);
if (isAvailableB1 && (!isAvailableA1 || !cuLeft->hasEqualMotion(leftPartIdx, *cuAbove, abovePartIdx)))
{
// get Inter Dir
candDir[count] = cuAbove->m_interDir[abovePartIdx];
// get Mv from Left
cuAbove->getMvField(cuAbove, abovePartIdx, 0, candMvField[count][0]);
if (isInterB)
cuAbove->getMvField(cuAbove, abovePartIdx, 1, candMvField[count][1]);
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
// above right
uint32_t aboveRightPartIdx = 0;
const CUData* cuAboveRight = getPUAboveRight(aboveRightPartIdx, partIdxRT);
bool isAvailableB0 = cuAboveRight &&
cuAboveRight->isDiffMER(xP + nPSW, yP - 1, xP, yP) &&
cuAboveRight->isInter(aboveRightPartIdx);
if (isAvailableB0 && (!isAvailableB1 || !cuAbove->hasEqualMotion(abovePartIdx, *cuAboveRight, aboveRightPartIdx)))
{
// get Inter Dir
candDir[count] = cuAboveRight->m_interDir[aboveRightPartIdx];
// get Mv from Left
cuAboveRight->getMvField(cuAboveRight, aboveRightPartIdx, 0, candMvField[count][0]);
if (isInterB)
cuAboveRight->getMvField(cuAboveRight, aboveRightPartIdx, 1, candMvField[count][1]);
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
// left bottom
uint32_t leftBottomPartIdx = 0;
const CUData* cuLeftBottom = this->getPUBelowLeft(leftBottomPartIdx, partIdxLB);
bool isAvailableA0 = cuLeftBottom &&
cuLeftBottom->isDiffMER(xP - 1, yP + nPSH, xP, yP) &&
cuLeftBottom->isInter(leftBottomPartIdx);
if (isAvailableA0 && (!isAvailableA1 || !cuLeft->hasEqualMotion(leftPartIdx, *cuLeftBottom, leftBottomPartIdx)))
{
// get Inter Dir
candDir[count] = cuLeftBottom->m_interDir[leftBottomPartIdx];
// get Mv from Left
cuLeftBottom->getMvField(cuLeftBottom, leftBottomPartIdx, 0, candMvField[count][0]);
if (isInterB)
cuLeftBottom->getMvField(cuLeftBottom, leftBottomPartIdx, 1, candMvField[count][1]);
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
// above left
if (count < 4)
{
uint32_t aboveLeftPartIdx = 0;
const CUData* cuAboveLeft = getPUAboveLeft(aboveLeftPartIdx, absPartAddr);
bool isAvailableB2 = cuAboveLeft &&
cuAboveLeft->isDiffMER(xP - 1, yP - 1, xP, yP) &&
cuAboveLeft->isInter(aboveLeftPartIdx);
if (isAvailableB2 && (!isAvailableA1 || !cuLeft->hasEqualMotion(leftPartIdx, *cuAboveLeft, aboveLeftPartIdx))
&& (!isAvailableB1 || !cuAbove->hasEqualMotion(abovePartIdx, *cuAboveLeft, aboveLeftPartIdx)))
{
// get Inter Dir
candDir[count] = cuAboveLeft->m_interDir[aboveLeftPartIdx];
// get Mv from Left
cuAboveLeft->getMvField(cuAboveLeft, aboveLeftPartIdx, 0, candMvField[count][0]);
if (isInterB)
cuAboveLeft->getMvField(cuAboveLeft, aboveLeftPartIdx, 1, candMvField[count][1]);
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
}
if (m_slice->m_sps->bTemporalMVPEnabled)
{
uint32_t partIdxRB = deriveRightBottomIdx(puIdx);
MV colmv;
int ctuIdx = -1;
// image boundary check
if (m_encData->getPicCTU(m_cuAddr)->m_cuPelX + g_zscanToPelX[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picWidthInLumaSamples &&
m_encData->getPicCTU(m_cuAddr)->m_cuPelY + g_zscanToPelY[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picHeightInLumaSamples)
{
uint32_t absPartIdxRB = g_zscanToRaster[partIdxRB];
uint32_t numUnits = s_numPartInCUSize;
bool bNotLastCol = lessThanCol(absPartIdxRB, numUnits - 1, numUnits); // is not at the last column of CTU
bool bNotLastRow = lessThanRow(absPartIdxRB, numUnits - 1, numUnits); // is not at the last row of CTU
if (bNotLastCol && bNotLastRow)
{
absPartAddr = g_rasterToZscan[absPartIdxRB + numUnits + 1];
ctuIdx = m_cuAddr;
}
else if (bNotLastCol)
absPartAddr = g_rasterToZscan[(absPartIdxRB + numUnits + 1) & (numUnits - 1)];
else if (bNotLastRow)
{
absPartAddr = g_rasterToZscan[absPartIdxRB + 1];
ctuIdx = m_cuAddr + 1;
}
else // is the right bottom corner of CTU
absPartAddr = 0;
}
int maxList = isInterB ? 2 : 1;
int dir = 0, refIdx = 0;
for (int list = 0; list < maxList; list++)
{
bool bExistMV = ctuIdx >= 0 && getColMVP(colmv, refIdx, list, ctuIdx, absPartAddr);
if (!bExistMV)
{
uint32_t partIdxCenter = deriveCenterIdx(puIdx);
bExistMV = getColMVP(colmv, refIdx, list, m_cuAddr, partIdxCenter);
}
if (bExistMV)
{
dir |= (1 << list);
candMvField[count][list].mv = colmv;
candMvField[count][list].refIdx = refIdx;
}
}
if (dir != 0)
{
candDir[count] = (uint8_t)dir;
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
}
if (isInterB)
{
const uint32_t cutoff = count * (count - 1);
uint32_t priorityList0 = 0xEDC984; // { 0, 1, 0, 2, 1, 2, 0, 3, 1, 3, 2, 3 }
uint32_t priorityList1 = 0xB73621; // { 1, 0, 2, 0, 2, 1, 3, 0, 3, 1, 3, 2 }
for (uint32_t idx = 0; idx < cutoff; idx++, priorityList0 >>= 2, priorityList1 >>= 2)
{
int i = priorityList0 & 3;
int j = priorityList1 & 3;
if ((candDir[i] & 0x1) && (candDir[j] & 0x2))
{
// get Mv from cand[i] and cand[j]
int refIdxL0 = candMvField[i][0].refIdx;
int refIdxL1 = candMvField[j][1].refIdx;
int refPOCL0 = m_slice->m_refPOCList[0][refIdxL0];
int refPOCL1 = m_slice->m_refPOCList[1][refIdxL1];
if (!(refPOCL0 == refPOCL1 && candMvField[i][0].mv == candMvField[j][1].mv))
{
candMvField[count][0].mv = candMvField[i][0].mv;
candMvField[count][0].refIdx = refIdxL0;
candMvField[count][1].mv = candMvField[j][1].mv;
candMvField[count][1].refIdx = refIdxL1;
candDir[count] = 3;
if (++count == maxNumMergeCand)
return maxNumMergeCand;
}
}
}
}
int numRefIdx = (isInterB) ? X265_MIN(m_slice->m_numRefIdx[0], m_slice->m_numRefIdx[1]) : m_slice->m_numRefIdx[0];
int r = 0;
int refcnt = 0;
while (count < maxNumMergeCand)
{
candDir[count] = 1;
candMvField[count][0].mv.word = 0;
candMvField[count][0].refIdx = r;
if (isInterB)
{
candDir[count] = 3;
candMvField[count][1].mv.word = 0;
candMvField[count][1].refIdx = r;
}
count++;
if (refcnt == numRefIdx - 1)
r = 0;
else
{
++r;
++refcnt;
}
}
return count;
}
// Create the PMV list. Called for each reference index.
int CUData::getPMV(InterNeighbourMV *neighbours, uint32_t picList, uint32_t refIdx, MV* amvpCand, MV* pmv) const
{
MV directMV[MD_ABOVE_LEFT + 1];
MV indirectMV[MD_ABOVE_LEFT + 1];
bool validDirect[MD_ABOVE_LEFT + 1];
bool validIndirect[MD_ABOVE_LEFT + 1];
// Left candidate.
validDirect[MD_BELOW_LEFT] = getDirectPMV(directMV[MD_BELOW_LEFT], neighbours + MD_BELOW_LEFT, picList, refIdx);
validDirect[MD_LEFT] = getDirectPMV(directMV[MD_LEFT], neighbours + MD_LEFT, picList, refIdx);
// Top candidate.
validDirect[MD_ABOVE_RIGHT] = getDirectPMV(directMV[MD_ABOVE_RIGHT], neighbours + MD_ABOVE_RIGHT, picList, refIdx);
validDirect[MD_ABOVE] = getDirectPMV(directMV[MD_ABOVE], neighbours + MD_ABOVE, picList, refIdx);
validDirect[MD_ABOVE_LEFT] = getDirectPMV(directMV[MD_ABOVE_LEFT], neighbours + MD_ABOVE_LEFT, picList, refIdx);
// Left candidate.
validIndirect[MD_BELOW_LEFT] = getIndirectPMV(indirectMV[MD_BELOW_LEFT], neighbours + MD_BELOW_LEFT, picList, refIdx);
validIndirect[MD_LEFT] = getIndirectPMV(indirectMV[MD_LEFT], neighbours + MD_LEFT, picList, refIdx);
// Top candidate.
validIndirect[MD_ABOVE_RIGHT] = getIndirectPMV(indirectMV[MD_ABOVE_RIGHT], neighbours + MD_ABOVE_RIGHT, picList, refIdx);
validIndirect[MD_ABOVE] = getIndirectPMV(indirectMV[MD_ABOVE], neighbours + MD_ABOVE, picList, refIdx);
validIndirect[MD_ABOVE_LEFT] = getIndirectPMV(indirectMV[MD_ABOVE_LEFT], neighbours + MD_ABOVE_LEFT, picList, refIdx);
int num = 0;
// Left predictor search
if (validDirect[MD_BELOW_LEFT])
amvpCand[num++] = directMV[MD_BELOW_LEFT];
else if (validDirect[MD_LEFT])
amvpCand[num++] = directMV[MD_LEFT];
else if (validIndirect[MD_BELOW_LEFT])
amvpCand[num++] = indirectMV[MD_BELOW_LEFT];
else if (validIndirect[MD_LEFT])
amvpCand[num++] = indirectMV[MD_LEFT];
bool bAddedSmvp = num > 0;
// Above predictor search
if (validDirect[MD_ABOVE_RIGHT])
amvpCand[num++] = directMV[MD_ABOVE_RIGHT];
else if (validDirect[MD_ABOVE])
amvpCand[num++] = directMV[MD_ABOVE];
else if (validDirect[MD_ABOVE_LEFT])
amvpCand[num++] = directMV[MD_ABOVE_LEFT];
if (!bAddedSmvp)
{
if (validIndirect[MD_ABOVE_RIGHT])
amvpCand[num++] = indirectMV[MD_ABOVE_RIGHT];
else if (validIndirect[MD_ABOVE])
amvpCand[num++] = indirectMV[MD_ABOVE];
else if (validIndirect[MD_ABOVE_LEFT])
amvpCand[num++] = indirectMV[MD_ABOVE_LEFT];
}
int numMvc = 0;
for (int dir = MD_LEFT; dir <= MD_ABOVE_LEFT; dir++)
{
if (validDirect[dir] && directMV[dir].notZero())
pmv[numMvc++] = directMV[dir];
if (validIndirect[dir] && indirectMV[dir].notZero())
pmv[numMvc++] = indirectMV[dir];
}
if (num == 2)
num -= amvpCand[0] == amvpCand[1];
// Get the collocated candidate. At this step, either the first candidate
// was found or its value is 0.
if (m_slice->m_sps->bTemporalMVPEnabled && num < 2)
{
int tempRefIdx = neighbours[MD_COLLOCATED].refIdx[picList];
if (tempRefIdx != -1)
{
uint32_t cuAddr = neighbours[MD_COLLOCATED].cuAddr[picList];
const Frame* colPic = m_slice->m_refPicList[m_slice->isInterB() && !m_slice->m_colFromL0Flag][m_slice->m_colRefIdx];
const CUData* colCU = colPic->m_encData->getPicCTU(cuAddr);
// Scale the vector
int colRefPOC = colCU->m_slice->m_refPOCList[tempRefIdx >> 4][tempRefIdx & 0xf];
int colPOC = colCU->m_slice->m_poc;
int curRefPOC = m_slice->m_refPOCList[picList][refIdx];
int curPOC = m_slice->m_poc;
pmv[numMvc++] = amvpCand[num++] = scaleMvByPOCDist(neighbours[MD_COLLOCATED].mv[picList], curPOC, curRefPOC, colPOC, colRefPOC);
}
}
while (num < AMVP_NUM_CANDS)
amvpCand[num++] = 0;
return numMvc;
}
/* Constructs a list of candidates for AMVP, and a larger list of motion candidates */
void CUData::getNeighbourMV(uint32_t puIdx, uint32_t absPartIdx, InterNeighbourMV* neighbours) const
{
// Set the temporal neighbour to unavailable by default.
neighbours[MD_COLLOCATED].unifiedRef = -1;
uint32_t partIdxLT, partIdxRT, partIdxLB = deriveLeftBottomIdx(puIdx);
deriveLeftRightTopIdx(puIdx, partIdxLT, partIdxRT);
// Load the spatial MVs.
getInterNeighbourMV(neighbours + MD_BELOW_LEFT, partIdxLB, MD_BELOW_LEFT);
getInterNeighbourMV(neighbours + MD_LEFT, partIdxLB, MD_LEFT);
getInterNeighbourMV(neighbours + MD_ABOVE_RIGHT,partIdxRT, MD_ABOVE_RIGHT);
getInterNeighbourMV(neighbours + MD_ABOVE, partIdxRT, MD_ABOVE);
getInterNeighbourMV(neighbours + MD_ABOVE_LEFT, partIdxLT, MD_ABOVE_LEFT);
if (m_slice->m_sps->bTemporalMVPEnabled)
{
uint32_t absPartAddr = m_absIdxInCTU + absPartIdx;
uint32_t partIdxRB = deriveRightBottomIdx(puIdx);
// co-located RightBottom temporal predictor (H)
int ctuIdx = -1;
// image boundary check
if (m_encData->getPicCTU(m_cuAddr)->m_cuPelX + g_zscanToPelX[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picWidthInLumaSamples &&
m_encData->getPicCTU(m_cuAddr)->m_cuPelY + g_zscanToPelY[partIdxRB] + UNIT_SIZE < m_slice->m_sps->picHeightInLumaSamples)
{
uint32_t absPartIdxRB = g_zscanToRaster[partIdxRB];
uint32_t numUnits = s_numPartInCUSize;
bool bNotLastCol = lessThanCol(absPartIdxRB, numUnits - 1, numUnits); // is not at the last column of CTU
bool bNotLastRow = lessThanRow(absPartIdxRB, numUnits - 1, numUnits); // is not at the last row of CTU
if (bNotLastCol && bNotLastRow)
{
absPartAddr = g_rasterToZscan[absPartIdxRB + numUnits + 1];
ctuIdx = m_cuAddr;
}
else if (bNotLastCol)
absPartAddr = g_rasterToZscan[(absPartIdxRB + numUnits + 1) & (numUnits - 1)];
else if (bNotLastRow)
{
absPartAddr = g_rasterToZscan[absPartIdxRB + 1];
ctuIdx = m_cuAddr + 1;
}
else // is the right bottom corner of CTU
absPartAddr = 0;
}
if (!(ctuIdx >= 0 && getCollocatedMV(ctuIdx, absPartAddr, neighbours + MD_COLLOCATED)))
{
uint32_t partIdxCenter = deriveCenterIdx(puIdx);
uint32_t curCTUIdx = m_cuAddr;
getCollocatedMV(curCTUIdx, partIdxCenter, neighbours + MD_COLLOCATED);
}
}
}
void CUData::getInterNeighbourMV(InterNeighbourMV *neighbour, uint32_t partUnitIdx, MVP_DIR dir) const
{
const CUData* tmpCU = NULL;
uint32_t idx = 0;
switch (dir)
{
case MD_LEFT:
tmpCU = getPULeft(idx, partUnitIdx);
break;
case MD_ABOVE:
tmpCU = getPUAbove(idx, partUnitIdx);
break;
case MD_ABOVE_RIGHT:
tmpCU = getPUAboveRight(idx, partUnitIdx);
break;
case MD_BELOW_LEFT:
tmpCU = getPUBelowLeft(idx, partUnitIdx);
break;
case MD_ABOVE_LEFT:
tmpCU = getPUAboveLeft(idx, partUnitIdx);
break;
default:
break;
}
if (!tmpCU)
{
// Mark the PMV as unavailable.
for (int i = 0; i < 2; i++)
neighbour->refIdx[i] = -1;
return;
}
for (int i = 0; i < 2; i++)
{
// Get the MV.
neighbour->mv[i] = tmpCU->m_mv[i][idx];
// Get the reference idx.
neighbour->refIdx[i] = tmpCU->m_refIdx[i][idx];
}
}
/* Clip motion vector to within slightly padded boundary of picture (the
* MV may reference a block that is completely within the padded area).
* Note this function is unaware of how much of this picture is actually
* available for use (re: frame parallelism) */
void CUData::clipMv(MV& outMV) const
{
const uint32_t mvshift = 2;
uint32_t offset = 8;
int16_t xmax = (int16_t)((m_slice->m_sps->picWidthInLumaSamples + offset - m_cuPelX - 1) << mvshift);
int16_t xmin = -(int16_t)((g_maxCUSize + offset + m_cuPelX - 1) << mvshift);
int16_t ymax = (int16_t)((m_slice->m_sps->picHeightInLumaSamples + offset - m_cuPelY - 1) << mvshift);
int16_t ymin = -(int16_t)((g_maxCUSize + offset + m_cuPelY - 1) << mvshift);
outMV.x = X265_MIN(xmax, X265_MAX(xmin, outMV.x));
outMV.y = X265_MIN(ymax, X265_MAX(ymin, outMV.y));
}
// Load direct spatial MV if available.
bool CUData::getDirectPMV(MV& pmv, InterNeighbourMV *neighbours, uint32_t picList, uint32_t refIdx) const
{
int curRefPOC = m_slice->m_refPOCList[picList][refIdx];
for (int i = 0; i < 2; i++, picList = !picList)
{
int partRefIdx = neighbours->refIdx[picList];
if (partRefIdx >= 0 && curRefPOC == m_slice->m_refPOCList[picList][partRefIdx])
{
pmv = neighbours->mv[picList];
return true;
}
}
return false;
}
// Load indirect spatial MV if available. An indirect MV has to be scaled.
bool CUData::getIndirectPMV(MV& outMV, InterNeighbourMV *neighbours, uint32_t picList, uint32_t refIdx) const
{
int curPOC = m_slice->m_poc;
int neibPOC = curPOC;
int curRefPOC = m_slice->m_refPOCList[picList][refIdx];
for (int i = 0; i < 2; i++, picList = !picList)
{
int partRefIdx = neighbours->refIdx[picList];
if (partRefIdx >= 0)
{
int neibRefPOC = m_slice->m_refPOCList[picList][partRefIdx];
MV mvp = neighbours->mv[picList];
outMV = scaleMvByPOCDist(mvp, curPOC, curRefPOC, neibPOC, neibRefPOC);
return true;
}
}
return false;
}
bool CUData::getColMVP(MV& outMV, int& outRefIdx, int picList, int cuAddr, int partUnitIdx) const
{
const Frame* colPic = m_slice->m_refPicList[m_slice->isInterB() && !m_slice->m_colFromL0Flag][m_slice->m_colRefIdx];
const CUData* colCU = colPic->m_encData->getPicCTU(cuAddr);
uint32_t absPartAddr = partUnitIdx & TMVP_UNIT_MASK;
if (colCU->m_predMode[partUnitIdx] == MODE_NONE || colCU->isIntra(absPartAddr))
return false;
int colRefPicList = m_slice->m_bCheckLDC ? picList : m_slice->m_colFromL0Flag;
int colRefIdx = colCU->m_refIdx[colRefPicList][absPartAddr];
if (colRefIdx < 0)
{
colRefPicList = !colRefPicList;
colRefIdx = colCU->m_refIdx[colRefPicList][absPartAddr];
if (colRefIdx < 0)
return false;
}
// Scale the vector
int colRefPOC = colCU->m_slice->m_refPOCList[colRefPicList][colRefIdx];
int colPOC = colCU->m_slice->m_poc;
MV colmv = colCU->m_mv[colRefPicList][absPartAddr];
int curRefPOC = m_slice->m_refPOCList[picList][outRefIdx];
int curPOC = m_slice->m_poc;
outMV = scaleMvByPOCDist(colmv, curPOC, curRefPOC, colPOC, colRefPOC);
return true;
}
// Cache the collocated MV.
bool CUData::getCollocatedMV(int cuAddr, int partUnitIdx, InterNeighbourMV *neighbour) const
{
const Frame* colPic = m_slice->m_refPicList[m_slice->isInterB() && !m_slice->m_colFromL0Flag][m_slice->m_colRefIdx];
const CUData* colCU = colPic->m_encData->getPicCTU(cuAddr);
uint32_t absPartAddr = partUnitIdx & TMVP_UNIT_MASK;
if (colCU->m_predMode[partUnitIdx] == MODE_NONE || colCU->isIntra(absPartAddr))
return false;
for (int list = 0; list < 2; list++)
{
neighbour->cuAddr[list] = cuAddr;
int colRefPicList = m_slice->m_bCheckLDC ? list : m_slice->m_colFromL0Flag;
int colRefIdx = colCU->m_refIdx[colRefPicList][absPartAddr];
if (colRefIdx < 0)
colRefPicList = !colRefPicList;
neighbour->refIdx[list] = colCU->m_refIdx[colRefPicList][absPartAddr];
neighbour->refIdx[list] |= colRefPicList << 4;
neighbour->mv[list] = colCU->m_mv[colRefPicList][absPartAddr];
}
return neighbour->unifiedRef != -1;
}
MV CUData::scaleMvByPOCDist(const MV& inMV, int curPOC, int curRefPOC, int colPOC, int colRefPOC) const
{
int diffPocD = colPOC - colRefPOC;
int diffPocB = curPOC - curRefPOC;
if (diffPocD == diffPocB)
return inMV;
else
{
int tdb = x265_clip3(-128, 127, diffPocB);
int tdd = x265_clip3(-128, 127, diffPocD);
int x = (0x4000 + abs(tdd / 2)) / tdd;
int scale = x265_clip3(-4096, 4095, (tdb * x + 32) >> 6);
return scaleMv(inMV, scale);
}
}
uint32_t CUData::deriveCenterIdx(uint32_t puIdx) const
{
uint32_t absPartIdx;
int puWidth, puHeight;
getPartIndexAndSize(puIdx, absPartIdx, puWidth, puHeight);
return g_rasterToZscan[g_zscanToRaster[m_absIdxInCTU + absPartIdx]
+ (puHeight >> (LOG2_UNIT_SIZE + 1)) * s_numPartInCUSize
+ (puWidth >> (LOG2_UNIT_SIZE + 1))];
}
/** 函数功能 ：？得到TU的熵编码参数，主要用于获取扫描方式。
** HEVC中对尺寸小的TU（Luma 8x8/4x4, Chroma 4x4/2x2），使用模式依赖的模式选择（Mode Dependent Coefficient Scanning, MDCS）
* \参数 result ：返回的熵编码参数
* \参数 absPartIdx ：当前PU在当前CU下的zigzag偏移地址
* \参数 log2TrSize ：TU尺寸
* \参数 bIsLuma ：当前分量是否为亮度
**/
void CUData::getTUEntropyCodingParameters(TUEntropyCodingParameters &result, uint32_t absPartIdx, uint32_t log2TrSize, bool bIsLuma) const
{
bool bIsIntra = isIntra(absPartIdx); // 判断是否是Intra预测
// set the group layout
result.log2TrSizeCG = log2TrSize - 2; // 得到TU按照CG为单位计算的尺寸
// set the scan orders
if (bIsIntra) // 如果是Intra
{
uint32_t dirMode;
if (bIsLuma) // 如果是Luma，直接得到Luma的预测方向
dirMode = m_lumaIntraDir[absPartIdx];
else // 否则，找到chroma的预测方向
{
dirMode = m_chromaIntraDir[absPartIdx]; // 得到chroma的预测方向
if (dirMode == DM_CHROMA_IDX) // 如果是DM模式(即直接使用Luma的预测方向)，则需要找到相应的Luma分量所选择的预测方向
{
dirMode = m_lumaIntraDir[(m_chromaFormat == X265_CSP_I444) ? absPartIdx : absPartIdx & 0xFC]; // 得到对应亮度块的预测方向，对Yuv444和Yuv422，需要得到其对应预测方向
dirMode = (m_chromaFormat == X265_CSP_I422) ? g_chroma422IntraAngleMappingTable[dirMode] : dirMode;
}
}
// 尺寸小的TU（Luma 8x8/4x4, Chroma 4x4/2x2）根据预测方向来选择扫描方式
// MDCS_LOG2_MAX_SIZE = 3，对于Luma分量m_hChromaShift=0，所以就是log2TrSize<=3（即TU size<=8x8）则可以进行MDCS的操作
// 对于Chroma分量m_hChromaShift=1，所以log2TrSize<=2（即TU size<=4x4）可以进行MDCS的操作
// MDCS根据预测方向选择扫描方式，如果预测方向是竖直方向或其临近方向则使用水平扫描，如果预测方向是水平方向或其临近方向则使用竖直扫描方式
if (log2TrSize <= (MDCS_LOG2_MAX_SIZE - m_hChromaShift) || (bIsLuma && log2TrSize == MDCS_LOG2_MAX_SIZE))
result.scanType = dirMode >= 22 && dirMode <= 30 ? SCAN_HOR : dirMode >= 6 && dirMode <= 14 ? SCAN_VER : SCAN_DIAG;
else // 如果不满足MDCS的条件，则使用对角扫描方式
result.scanType = SCAN_DIAG;
}
else // inter预测模式都使用对角扫描方式
result.scanType = SCAN_DIAG;
result.scan = g_scanOrder[result.scanType][log2TrSize - 2]; // 根据扫描类型和TU尺寸得到系数的扫描顺序表
result.scanCG = g_scanOrderCG[result.scanType][result.log2TrSizeCG]; // 根据扫描类型和TU尺寸得到系数组CG的扫描顺序表
// 得到第一个非零系数的context下标
if (log2TrSize == 2)
result.firstSignificanceMapContext = 0;
else if (log2TrSize == 3)
result.firstSignificanceMapContext = (result.scanType != SCAN_DIAG && bIsLuma) ? 15 : 9;
else
result.firstSignificanceMapContext = bIsLuma ? 21 : 12;
}
#define CU_SET_FLAG(bitfield, flag, value) (bitfield) = ((bitfield) & (~(flag))) | ((~((value) - 1)) & (flag)) //设置geom falg 第一步清除当前标记位 第二步 设置当前标记位
/** 函数功能 ： 计算CU的几何信息
/* 调用范围 ： 只在FrameEncoder::initializeGeoms()函数中被调用
* \参数 ctuWidth ： 宽度的余数
* \参数 ctuHeight ： 高度的余数
* \参数 maxCUSize ： 最大CU
* \参数 minCUSize ： 最小CU
* \参数 cuDataArray： 存储位置
* 返回值 ： null
**/
void CUData::calcCTUGeoms(uint32_t ctuWidth, uint32_t ctuHeight, uint32_t maxCUSize, uint32_t minCUSize, CUGeom cuDataArray[CUGeom::MAX_GEOMS])
{
// Initialize the coding blocks inside the CTB
for (uint32_t log2CUSize = g_log2Size[maxCUSize], rangeCUIdx = 0; log2CUSize >= g_log2Size[minCUSize]; log2CUSize--) //从最大CU遍历到最小CU
{
uint32_t blockSize = 1 << log2CUSize; //当前的块大小
uint32_t sbWidth = 1 << (g_log2Size[maxCUSize] - log2CUSize);//最大CU下有宽度上有几个当前块大小 如：64 ：1 32 ：2 16:4 8:8
int32_t lastLevelFlag = log2CUSize == g_log2Size[minCUSize];//判断当前位置是否为最小CU
for (uint32_t sbY = 0; sbY < sbWidth; sbY++) //按行遍历
{
for (uint32_t sbX = 0; sbX < sbWidth; sbX++)//遍历当前行的每个块
{
uint32_t depthIdx = g_depthScanIdx[sbY][sbX];//获取当前的zigzag号
uint32_t cuIdx = rangeCUIdx + depthIdx;//当前在geom的存储位置
uint32_t childIdx = rangeCUIdx + sbWidth * sbWidth + (depthIdx << 2);//对应位置子块在geom的存储位置
uint32_t px = sbX * blockSize;//在CTU中的pixel地址
uint32_t py = sbY * blockSize;//在CTU中的pixel地址
int32_t presentFlag = px < ctuWidth && py < ctuHeight;//判断当前块左上角像素是否在图像内部
int32_t splitMandatoryFlag = presentFlag && !lastLevelFlag && (px + blockSize > ctuWidth || py + blockSize > ctuHeight);//在图像内部 不是最小CU CU超过边界
/* Offset of the luma CU in the X, Y direction in terms of pixels from the CTU origin */
uint32_t xOffset = (sbX * blockSize) >> 3;
uint32_t yOffset = (sbY * blockSize) >> 3;
X265_CHECK(cuIdx < CUGeom::MAX_GEOMS, "CU geom index bug\n");
CUGeom *cu = cuDataArray + cuIdx;//获取存储位置
cu->log2CUSize = log2CUSize;//记录当前块大小
cu->childOffset = childIdx - cuIdx;//示从当前位置到第一个子cU的偏移量
cu->absPartIdx = g_depthScanIdx[yOffset][xOffset] * 4;// 当前CU在LCU中4x4 zizag地址
cu->numPartitions = (NUM_4x4_PARTITIONS >> ((g_maxLog2CUSize - cu->log2CUSize) * 2)); // 当前CU有多少4x4块
cu->depth = g_log2Size[maxCUSize] - log2CUSize;// 当前CU的深度
cu->flags = 0;
CU_SET_FLAG(cu->flags, CUGeom::PRESENT, presentFlag);//记录是否有PRESENT
CU_SET_FLAG(cu->flags, CUGeom::SPLIT_MANDATORY | CUGeom::SPLIT, splitMandatoryFlag);//记录 SPLIT_MANDATORY SPLIT
CU_SET_FLAG(cu->flags, CUGeom::LEAF, lastLevelFlag);//记录LEAF
}
}
rangeCUIdx += sbWidth * sbWidth;//用于计算每个块在geom的位置
}
}
扫一扫
1524
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
