/work/vvenc/source/Lib/EncoderLib/IntraSearch.cpp

Source
/* -----------------------------------------------------------------------------
The copyright in this software is being made available under the Clear BSD
License, included below. No patent rights, trademark rights and/or 
other Intellectual Property Rights other than the copyrights concerning 
the Software are granted under this license.

The Clear BSD License

Copyright (c) 2019-2026, Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. & The VVenC Authors.
All rights reserved.

Redistribution and use in source and binary forms, with or without modification,
are permitted (subject to the limitations in the disclaimer below) provided that
the following conditions are met:

     * Redistributions of source code must retain the above copyright notice,
     this list of conditions and the following disclaimer.

     * Redistributions in binary form must reproduce the above copyright
     notice, this list of conditions and the following disclaimer in the
     documentation and/or other materials provided with the distribution.

     * Neither the name of the copyright holder nor the names of its
     contributors may be used to endorse or promote products derived from this
     software without specific prior written permission.

NO EXPRESS OR IMPLIED LICENSES TO ANY PARTY'S PATENT RIGHTS ARE GRANTED BY
THIS LICENSE. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.


------------------------------------------------------------------------------------------- */


/** \file     EncSearch.cpp
 *  \brief    encoder intra search class
 */

#include "IntraSearch.h"
#include "EncPicture.h"
#include "CommonLib/CommonDef.h"
#include "CommonLib/Rom.h"
#include "CommonLib/Picture.h"
#include "CommonLib/UnitTools.h"
#include "CommonLib/dtrace_next.h"
#include "CommonLib/dtrace_buffer.h"
#include <math.h>
#include "vvenc/vvencCfg.h"

//! \ingroup EncoderLib
//! \{

namespace vvenc {

#define PLTCtx(c) SubCtx( Ctx::Palette, c )

IntraSearch::IntraSearch()
  : m_pSaveCS       (nullptr)
  , m_pcEncCfg      (nullptr)
  , m_pcTrQuant     (nullptr)
  , m_pcRdCost      (nullptr)
  , m_CABACEstimator(nullptr)
  , m_CtxCache      (nullptr)
{
}

void IntraSearch::init(const VVEncCfg &encCfg, TrQuant *pTrQuant, RdCost *pRdCost, SortedPelUnitBufs<SORTED_BUFS> *pSortedPelUnitBufs, XUCache &unitCache )
{
  IntraPrediction::init( encCfg.m_internChromaFormat, encCfg.m_internalBitDepth[ CH_L ] );

  m_pcEncCfg          = &encCfg;
  m_pcTrQuant         = pTrQuant;
  m_pcRdCost          = pRdCost;
  m_SortedPelUnitBufs = pSortedPelUnitBufs;

  const ChromaFormat chrFormat = encCfg.m_internChromaFormat;
  const int maxCUSize          = encCfg.m_CTUSize;

  Area area = Area( 0, 0, maxCUSize, maxCUSize );

  m_pTempCS = new CodingStructure( unitCache, nullptr );
  m_pBestCS = new CodingStructure( unitCache, nullptr );

  m_pTempCS->createForSearch( chrFormat, area );
  m_pBestCS->createForSearch( chrFormat, area );

  const int uiNumSaveLayersToAllocate = 3;
  m_pSaveCS = new CodingStructure*[uiNumSaveLayersToAllocate];
  for( int layer = 0; layer < uiNumSaveLayersToAllocate; layer++ )
  {
    m_pSaveCS[ layer ] = new CodingStructure( unitCache, nullptr );
    m_pSaveCS[ layer ]->createForSearch( chrFormat, Area( 0, 0, maxCUSize, maxCUSize ) );
    m_pSaveCS[ layer ]->initStructData();
  }

  CompArea chromaArea( COMP_Cb, chrFormat, area, true );
  for( int i = 0; i < 5; i++ )
  {
    m_orgResiCb[i].create( chromaArea );
    m_orgResiCr[i].create( chromaArea );
  }
}

void IntraSearch::destroy()
{
  if ( m_pSaveCS )
  {
    const int uiNumSaveLayersToAllocate = 3;
    for( int layer = 0; layer < uiNumSaveLayersToAllocate; layer++ )
    {
      if ( m_pSaveCS[ layer ] ) { m_pSaveCS[ layer ]->destroy(); delete m_pSaveCS[ layer ]; }
    }
    delete[] m_pSaveCS;
    m_pSaveCS = nullptr;
  }

  if( m_pTempCS )
  {
    m_pTempCS->destroy();
    delete m_pTempCS; m_pTempCS = nullptr;
  }

  if( m_pBestCS )
  {
    m_pBestCS->destroy();
    delete m_pBestCS; m_pBestCS = nullptr;
  }
}

IntraSearch::~IntraSearch()
{
  destroy();
}

void IntraSearch::setCtuEncRsrc( CABACWriter* cabacEstimator, CtxCache *ctxCache )
{
  m_CABACEstimator = cabacEstimator;
  m_CtxCache       = ctxCache;
}

//////////////////////////////////////////////////////////////////////////
// INTRA PREDICTION
//////////////////////////////////////////////////////////////////////////
static constexpr double COST_UNKNOWN = -65536.0;

double IntraSearch::xFindInterCUCost( CodingUnit &cu )
{
  if( CU::isConsIntra(cu) && !cu.slice->isIntra() )
  {
    //search corresponding inter CU cost
    for( int i = 0; i < m_numCuInSCIPU; i++ )
    {
      if( cu.lumaPos() == m_cuAreaInSCIPU[i].pos() && cu.lumaSize() == m_cuAreaInSCIPU[i].size() )
      {
        return m_cuCostInSCIPU[i];
      }
    }
  }
  return COST_UNKNOWN;
}

void IntraSearch::xEstimateLumaRdModeList(int& numModesForFullRD,
  static_vector<ModeInfo, FAST_UDI_MAX_RDMODE_NUM>& RdModeList,
  static_vector<ModeInfo, FAST_UDI_MAX_RDMODE_NUM>& HadModeList,
  static_vector<double, FAST_UDI_MAX_RDMODE_NUM>& CandCostList,
  static_vector<double, FAST_UDI_MAX_RDMODE_NUM>& CandHadList, CodingUnit& cu, bool testMip )
{
  PROFILER_SCOPE_AND_STAGE_EXT( 1, _TPROF, P_INTRA_EST_RD_CAND, cu.cs, CH_L );
  const uint16_t intra_ctx_size = Ctx::IntraLumaMpmFlag.size() + Ctx::IntraLumaPlanarFlag.size() + Ctx::MultiRefLineIdx.size() + Ctx::ISPMode.size() + Ctx::MipFlag.size();
  const TempCtx  ctxStartIntraCtx(m_CtxCache, SubCtx(CtxSet(Ctx::IntraLumaMpmFlag(), intra_ctx_size), m_CABACEstimator->getCtx()));
  const double   sqrtLambdaForFirstPass = m_pcRdCost->getMotionLambda() * FRAC_BITS_SCALE;
  const int numModesAvailable = NUM_LUMA_MODE; // total number of Intra modes

  CHECK(numModesForFullRD >= numModesAvailable, "Too many modes for full RD search");

  const SPS& sps     = *cu.cs->sps;
  const bool fastMip = sps.MIP && m_pcEncCfg->m_useFastMIP;

  // this should always be true
  CHECK( !cu.Y().valid(), "CU is not valid" );

  const CompArea& area = cu.Y();

  const UnitArea localUnitArea(area.chromaFormat, Area(0, 0, area.width, area.height));
  if( testMip)
  {
    numModesForFullRD += fastMip ? numModesForFullRD - std::min( m_pcEncCfg->m_useFastMIP, numModesForFullRD )
                                 : numModesForFullRD;
    m_SortedPelUnitBufs->prepare( localUnitArea, numModesForFullRD + 1 );
  }
  else
  {
    m_SortedPelUnitBufs->prepare( localUnitArea, numModesForFullRD );
  }

  CPelBuf piOrg   = cu.cs->getOrgBuf(COMP_Y);
  PelBuf piPred  = m_SortedPelUnitBufs->getTestBuf(COMP_Y);

  DistParam distParam    = m_pcRdCost->setDistParam( piOrg, piPred, sps.bitDepths[ CH_L ], DF_HAD_2SAD); // Use HAD (SATD) cost

  const int numHadCand = (testMip ? 2 : 1) * 3;

  //*** Derive (regular) candidates using Hadamard
  cu.mipFlag = false;
  cu.multiRefIdx = 0;

  //===== init pattern for luma prediction =====
  initIntraPatternChType(cu, cu.Y(), true);

  bool satdChecked[NUM_INTRA_MODE] = { false };

  unsigned mpmLst[NUM_MOST_PROBABLE_MODES];
  CU::getIntraMPMs(cu, mpmLst);

  const int decMsk = ( 1 << m_pcEncCfg->m_IntraEstDecBit ) - 1;

  m_parentCandList.resize( 0 );
  m_parentCandList.reserve( ( numModesAvailable >> m_pcEncCfg->m_IntraEstDecBit ) + 2 );

  for( unsigned mode = 0; mode < numModesAvailable; mode++ )
  {
    // Skip checking extended Angular modes in the first round of SATD
    if( mode > DC_IDX && ( mode & decMsk ) )
    {
      continue;
    }

    m_parentCandList.push_back( ModeInfo( false, false, 0, NOT_INTRA_SUBPARTITIONS, mode ) );
  }
   
  for( int decDst = 1 << m_pcEncCfg->m_IntraEstDecBit; decDst > 0; decDst >>= 1 )
  {
    for( unsigned idx = 0; idx < m_parentCandList.size(); idx++ )
    {
      int modeParent = m_parentCandList[idx].modeId;

      int off = decDst & decMsk;
      int inc = decDst << 1;

#if 1 // INTRA_AS_IN_VTM
      if( off != 0 && ( modeParent <= ( DC_IDX + 1 ) || modeParent >= ( NUM_LUMA_MODE - 1 ) ) )
      {
        continue;
      }

#endif
      for( int mode = modeParent - off; mode < modeParent + off + 1; mode += inc )
      {
        if( satdChecked[mode] || mode < 0 || mode >= NUM_LUMA_MODE )
        {
          continue;
        }

        cu.intraDir[0] = mode;

        initPredIntraParams( cu, cu.Y(), sps );
        distParam.cur.buf = piPred.buf = m_SortedPelUnitBufs->getTestBuf().Y().buf;
        predIntraAng( COMP_Y, piPred, cu );

        // Use the min between SAD and HAD as the cost criterion
        // SAD is scaled by 2 to align with the scaling of HAD
        Distortion minSadHad = distParam.distFunc( distParam );

        uint64_t fracModeBits = xFracModeBitsIntraLuma( cu, mpmLst );

        //restore ctx
        m_CABACEstimator->getCtx() = SubCtx( CtxSet( Ctx::IntraLumaMpmFlag(), intra_ctx_size ), ctxStartIntraCtx );

        double cost = ( double ) minSadHad + ( double ) fracModeBits * sqrtLambdaForFirstPass;
        DTRACE( g_trace_ctx, D_INTRA_COST, "IntraHAD: %u, %llu, %f (%d)\n", minSadHad, fracModeBits, cost, mode );

        int insertPos = -1;
        updateCandList( ModeInfo( false, false, 0, NOT_INTRA_SUBPARTITIONS, mode ), cost, RdModeList, CandCostList, numModesForFullRD, &insertPos );
        updateCandList( ModeInfo( false, false, 0, NOT_INTRA_SUBPARTITIONS, mode ), ( double ) minSadHad, HadModeList, CandHadList, numHadCand );
        m_SortedPelUnitBufs->insert( insertPos, ( int ) RdModeList.size() );

        satdChecked[mode] = true;
      }
    }

    m_parentCandList.resize( RdModeList.size() );
    std::copy( RdModeList.cbegin(), RdModeList.cend(), m_parentCandList.begin() );
  }

  const bool isFirstLineOfCtu = (((cu.block(COMP_Y).y)&((cu.cs->sps)->CTUSize - 1)) == 0);
  if( m_pcEncCfg->m_MRL && ! isFirstLineOfCtu )
  {
    cu.multiRefIdx = 1;
    unsigned  multiRefMPM [NUM_MOST_PROBABLE_MODES];
    CU::getIntraMPMs(cu, multiRefMPM);

    for (int mRefNum = 1; mRefNum < MRL_NUM_REF_LINES; mRefNum++)
    {
      int multiRefIdx = MULTI_REF_LINE_IDX[mRefNum];

      cu.multiRefIdx = multiRefIdx;
      initIntraPatternChType(cu, cu.Y(), true);

      for (int x = 1; x < NUM_MOST_PROBABLE_MODES; x++)
      {
        cu.intraDir[0] = multiRefMPM[x];
        initPredIntraParams(cu, cu.Y(), sps);
        distParam.cur.buf = piPred.buf = m_SortedPelUnitBufs->getTestBuf().Y().buf;
        predIntraAng(COMP_Y, piPred, cu);

        // Use the min between SAD and SATD as the cost criterion
        // SAD is scaled by 2 to align with the scaling of HAD
        Distortion minSadHad = distParam.distFunc(distParam);

        // NB xFracModeBitsIntra will not affect the mode for chroma that may have already been pre-estimated.
        uint64_t fracModeBits = xFracModeBitsIntraLuma( cu, mpmLst );

        //restore ctx
        m_CABACEstimator->getCtx() = SubCtx(CtxSet(Ctx::IntraLumaMpmFlag(), intra_ctx_size), ctxStartIntraCtx);

        double cost = (double) minSadHad + (double) fracModeBits * sqrtLambdaForFirstPass;
//        DTRACE(g_trace_ctx, D_INTRA_COST, "IntraMRL: %u, %llu, %f (%d)\n", minSadHad, fracModeBits, cost, cu.intraDir[0]);

        int insertPos = -1;
        updateCandList( ModeInfo( false, false, multiRefIdx, NOT_INTRA_SUBPARTITIONS, cu.intraDir[0] ), cost, RdModeList,  CandCostList, numModesForFullRD, &insertPos );
        updateCandList( ModeInfo( false, false, multiRefIdx, NOT_INTRA_SUBPARTITIONS, cu.intraDir[0] ), (double)minSadHad, HadModeList, CandHadList,  numHadCand );
        m_SortedPelUnitBufs->insert(insertPos, (int)RdModeList.size());
      }
    }
    cu.multiRefIdx = 0;
  }

  if (testMip)
  {
    cu.mipFlag = true;
    cu.multiRefIdx = 0;

    double mipHadCost[MAX_NUM_MIP_MODE] = { MAX_DOUBLE };

    initIntraPatternChType(cu, cu.Y());
    initIntraMip( cu );

    const int transpOff    = getNumModesMip( cu.Y() );
    const int numModesFull = (transpOff << 1);
    for( uint32_t uiModeFull = 0; uiModeFull < numModesFull; uiModeFull++ )
    {
      const bool     isTransposed = (uiModeFull >= transpOff ? true : false);
      const uint32_t uiMode       = (isTransposed ? uiModeFull - transpOff : uiModeFull);

      cu.mipTransposedFlag = isTransposed;
      cu.intraDir[CH_L] = uiMode;
      distParam.cur.buf = piPred.buf = m_SortedPelUnitBufs->getTestBuf().Y().buf;
      predIntraMip(piPred, cu);

      // Use the min between SAD and HAD as the cost criterion
      // SAD is scaled by 2 to align with the scaling of HAD
      Distortion minSadHad = distParam.distFunc(distParam);

      uint64_t fracModeBits = xFracModeBitsIntraLuma( cu, mpmLst );

      //restore ctx
      m_CABACEstimator->getCtx() = SubCtx(CtxSet(Ctx::IntraLumaMpmFlag(), intra_ctx_size), ctxStartIntraCtx);

      double cost = double(minSadHad) + double(fracModeBits) * sqrtLambdaForFirstPass;
      mipHadCost[uiModeFull] = cost;
      DTRACE(g_trace_ctx, D_INTRA_COST, "IntraMIP: %u, %llu, %f (%d)\n", minSadHad, fracModeBits, cost, uiModeFull);

      int insertPos = -1;
      updateCandList( ModeInfo( true, isTransposed, 0, NOT_INTRA_SUBPARTITIONS, cu.intraDir[0] ), cost, RdModeList,  CandCostList, numModesForFullRD+1, &insertPos );
      updateCandList( ModeInfo( true, isTransposed, 0, NOT_INTRA_SUBPARTITIONS, cu.intraDir[0] ), 0.8*(double)minSadHad, HadModeList, CandHadList,  numHadCand );
      m_SortedPelUnitBufs->insert(insertPos, (int)RdModeList.size());
    }

    const double thresholdHadCost = 1.0 + 1.4 / sqrt((double)(cu.lwidth()*cu.lheight()));
    xReduceHadCandList(RdModeList, CandCostList, *m_SortedPelUnitBufs, numModesForFullRD, thresholdHadCost, mipHadCost, cu, fastMip);
  }

  if( m_pcEncCfg->m_bFastUDIUseMPMEnabled )
  {
    const int numMPMs = NUM_MOST_PROBABLE_MODES;
    unsigned  intraMpms[numMPMs];

    cu.multiRefIdx = 0;

    const int numCand = CU::getIntraMPMs( cu, intraMpms );
    ModeInfo mostProbableMode(false, false, 0, NOT_INTRA_SUBPARTITIONS, 0);

    for( int j = 0; j < numCand; j++ )
    {
      bool mostProbableModeIncluded = false;
      mostProbableMode.modeId = intraMpms[j];

      for( int i = 0; i < numModesForFullRD; i++ )
      {
        mostProbableModeIncluded |= ( mostProbableMode == RdModeList[i] );
      }
      if( !mostProbableModeIncluded )
      {
        numModesForFullRD++;
        RdModeList.push_back( mostProbableMode );
        CandCostList.push_back(0);
      }
    }
  }
}

bool IntraSearch::estIntraPredLumaQT(CodingUnit &cu, Partitioner &partitioner, double bestCost)
{
  CodingStructure       &cs           = *cu.cs;
  const int             width         = partitioner.currArea().lwidth();
  const int             height        = partitioner.currArea().lheight();

  //===== loop over partitions =====

  const TempCtx ctxStart           ( m_CtxCache, m_CABACEstimator->getCtx() );

  // variables for saving fast intra modes scan results across multiple LFNST passes
  double costInterCU = xFindInterCUCost( cu );

  bool validReturn = false;

  //===== determine set of modes to be tested (using prediction signal only) =====
  int numModesAvailable = NUM_LUMA_MODE; // total number of Intra modes
  static_vector<ModeInfo, FAST_UDI_MAX_RDMODE_NUM> RdModeList;
  static_vector<ModeInfo, FAST_UDI_MAX_RDMODE_NUM> HadModeList;
  static_vector<double, FAST_UDI_MAX_RDMODE_NUM> CandCostList;
  static_vector<double, FAST_UDI_MAX_RDMODE_NUM> CandHadList;

  int numModesForFullRD = g_aucIntraModeNumFast_UseMPM_2D[Log2(width) - MIN_CU_LOG2][Log2(height) - MIN_CU_LOG2];
  if (m_pcEncCfg->m_numIntraModesFullRD > 0)
    numModesForFullRD=m_pcEncCfg->m_numIntraModesFullRD;

#if INTRA_FULL_SEARCH
  numModesForFullRD = numModesAvailable;
#endif
  const SPS& sps = *cu.cs->sps;
  const bool mipAllowed = sps.MIP && cu.lwidth() <= sps.getMaxTbSize() && cu.lheight() <= sps.getMaxTbSize() && ((cu.lfnstIdx == 0) || allowLfnstWithMip(cu.lumaSize()));
  const int SizeThr     = 8 >> std::max( 0, m_pcEncCfg->m_useFastMIP - 1 );
  const bool testMip    = mipAllowed && ( cu.lwidth() <= ( SizeThr * cu.lheight() ) && cu.lheight() <= ( SizeThr * cu.lwidth() ) ) && ( cu.lwidth() <= MIP_MAX_WIDTH && cu.lheight() <= MIP_MAX_HEIGHT );
  bool testISP = sps.ISP && CU::canUseISP(width, height, cu.cs->sps->getMaxTbSize());
  if (testISP)
  {
    int numTotalPartsHor = (int)width >> floorLog2(CU::getISPSplitDim(width, height, TU_1D_VERT_SPLIT));
    int numTotalPartsVer = (int)height >> floorLog2(CU::getISPSplitDim(width, height, TU_1D_HORZ_SPLIT));
    m_ispTestedModes[0].init(numTotalPartsHor, numTotalPartsVer, 0);
    // the total number of subpartitions is modified to take into account the cases where LFNST cannot be combined with
    // ISP due to size restrictions
    numTotalPartsHor = sps.LFNST && CU::canUseLfnstWithISP(cu.Y(), HOR_INTRA_SUBPARTITIONS) ? numTotalPartsHor : 0;
    numTotalPartsVer = sps.LFNST && CU::canUseLfnstWithISP(cu.Y(), VER_INTRA_SUBPARTITIONS) ? numTotalPartsVer : 0;
    for (int j = 1; j < NUM_LFNST_NUM_PER_SET; j++)
    {
      m_ispTestedModes[j].init(numTotalPartsHor, numTotalPartsVer, 0);
    }
    testISP = m_ispTestedModes[0].numTotalParts[0];
  }
  else
  {
    m_ispTestedModes[0].init(0, 0, 0);
  }

  xEstimateLumaRdModeList(numModesForFullRD, RdModeList, HadModeList, CandCostList, CandHadList, cu, testMip);

  CHECK( (size_t)numModesForFullRD != RdModeList.size(), "Inconsistent state!" );

  // after this point, don't use numModesForFullRD
  if( m_pcEncCfg->m_usePbIntraFast && !cs.slice->isIntra() && RdModeList.size() < numModesAvailable )
  {
    double pbintraRatio = m_pcEncCfg->m_usePbIntraFast == 1 && ( cs.area.lwidth() >= 16 && cs.area.lheight() >= 16 ) ? 1.2 : PBINTRA_RATIO;

    int maxSize = -1;
    ModeInfo bestMipMode;
    int bestMipIdx = -1;
    for( int idx = 0; idx < RdModeList.size(); idx++ )
    {
      if( RdModeList[idx].mipFlg )
      {
        bestMipMode = RdModeList[idx];
        bestMipIdx = idx;
        break;
      }
    }
    const int numHadCand = 3;
    for (int k = numHadCand - 1; k >= 0; k--)
    {
      if (CandHadList.size() < (k + 1) || CandHadList[k] > cs.interHad * pbintraRatio) { maxSize = k; }
    }
    if (maxSize > 0)
    {
      RdModeList.resize(std::min<size_t>(RdModeList.size(), maxSize));
      if( bestMipIdx >= 0 )
      {
        if( RdModeList.size() <= bestMipIdx )
        {
          RdModeList.push_back(bestMipMode);
          m_SortedPelUnitBufs->swap( maxSize, bestMipIdx );
        }
      }
    }
    if (maxSize == 0)
    {
      cs.dist = MAX_DISTORTION;
      cs.interHad = 0;
      return false;
    }
  }

  //===== check modes (using r-d costs) =====
  ModeInfo bestPUMode;

  CodingStructure *csTemp = m_pTempCS;
  CodingStructure *csBest = m_pBestCS;

  csTemp->slice   = csBest->slice   = cs.slice;
  csTemp->picture = csBest->picture = cs.picture;
  csTemp->compactResize( cu );
  csBest->compactResize( cu );
  csTemp->initStructData();
  csBest->initStructData();

  int   bestLfnstIdx  = 0;
  const bool useBDPCM = cs.picture->useBDPCM;
  int   NumBDPCMCand  = (useBDPCM && sps.BDPCM && CU::bdpcmAllowed(cu, ComponentID(partitioner.chType))) ? 2 : 0;
  int   bestbdpcmMode = 0;
  int   bestISP       = 0;
  int   bestMrl       = 0;
  bool  bestMip       = 0;
  int   EndMode       = (int)RdModeList.size();
  bool  useISPlfnst   = testISP && sps.LFNST;
  bool  noLFNST_ts    = false;
  double bestCostIsp[2] = { MAX_DOUBLE, MAX_DOUBLE };
  bool disableMTS = false;
  bool disableLFNST = false;
  bool disableDCT2test = false;
  if (m_pcEncCfg->m_FastIntraTools)
  {
    int speedIntra = 0;
    xSpeedUpIntra(bestCost, EndMode, speedIntra, cu);
    disableMTS = (speedIntra >> 2 ) & 0x1;
    disableLFNST = (speedIntra >> 1) & 0x1;
    disableDCT2test = speedIntra>>3;
    if (disableLFNST)
    {
      noLFNST_ts = true;
      useISPlfnst = false;
    }
    if (speedIntra & 0x1)
    {
      testISP = false;
    }
  }

  for (int mode_cur = 0; mode_cur < EndMode + NumBDPCMCand; mode_cur++)
  {
    int mode = mode_cur;
    if (mode_cur >= EndMode)
    {
      mode = mode_cur - EndMode ? -1 : -2;
      testISP = false;
    }
    // set CU/PU to luma prediction mode
    ModeInfo testMode;
    int noISP = 0;
    int endISP = testISP ? 2 : 0;
    bool noLFNST = false || noLFNST_ts;
    if (mode && useISPlfnst)
    {
      noLFNST |= (bestCostIsp[0] > (bestCostIsp[1] * 1.4));
      if (mode > 2)
      {
        endISP = 0;
        testISP = false;
      }
    }
    if (testISP)
    {
      xSpeedUpISP(1, testISP, mode, noISP, endISP, cu, RdModeList, bestPUMode, bestISP, bestLfnstIdx);
    }
    int startISP = 0;
    if (disableDCT2test && mode && bestISP)
    {
      startISP = endISP ? 1 : 0;
    }
    for (int ispM = startISP; ispM <= endISP; ispM++)
    {
      if (ispM && (ispM == noISP))
      {
        continue;
      }

      if (mode < 0)
      {
        cu.bdpcmM[CH_L] = -mode;
        testMode = ModeInfo(false, false, 0, NOT_INTRA_SUBPARTITIONS, cu.bdpcmM[CH_L] == 2 ? VER_IDX : HOR_IDX);
      }
      else
      {
        testMode = RdModeList[mode];
        cu.bdpcmM[CH_L] = 0;
      }

      cu.ispMode = ispM;
      cu.mipFlag = testMode.mipFlg;
      cu.mipTransposedFlag = testMode.mipTrFlg;
      cu.multiRefIdx = testMode.mRefId;
      cu.intraDir[CH_L] = testMode.modeId;
      if (cu.ispMode && xSpeedUpISP(0, testISP, mode, noISP, endISP, cu, RdModeList, bestPUMode, bestISP, 0) )
      {
        continue;
      }
      if (m_pcEncCfg->m_FastIntraTools && (cu.ispMode || sps.LFNST || sps.MTS))
      {
        m_ispTestedModes[0].intraWasTested = true;
      }
      CHECK(cu.mipFlag && cu.multiRefIdx, "Error: combination of MIP and MRL not supported");
      CHECK(cu.multiRefIdx && (cu.intraDir[0] == PLANAR_IDX), "Error: combination of MRL and Planar mode not supported");
      CHECK(cu.ispMode && cu.mipFlag, "Error: combination of ISP and MIP not supported");
      CHECK(cu.ispMode && cu.multiRefIdx, "Error: combination of ISP and MRL not supported");

      // determine residual for partition
      cs.initSubStructure(*csTemp, partitioner.chType, cs.area, true);
      int doISP = (((cu.ispMode == 0) && noLFNST) || (useISPlfnst && mode && cu.ispMode && (bestLfnstIdx == 0)) || disableLFNST) ? -mode : mode;
      xIntraCodingLumaQT(*csTemp, partitioner, m_SortedPelUnitBufs->getBufFromSortedList(mode), bestCost, doISP, disableMTS);

      DTRACE(g_trace_ctx, D_INTRA_COST, "IntraCost T [x=%d,y=%d,w=%d,h=%d] %f (%d,%d,%d,%d,%d,%d) \n", cu.blocks[0].x,
        cu.blocks[0].y, width, height, csTemp->cost, testMode.modeId, testMode.ispMod,
        cu.multiRefIdx, cu.mipFlag, cu.lfnstIdx, cu.mtsFlag);

      if (cu.ispMode && !csTemp->cus[0]->firstTU->cbf[COMP_Y])
      {
        csTemp->cost = MAX_DOUBLE;
        csTemp->costDbOffset = 0;
      }
      if (useISPlfnst)
      {
        int n = (cu.ispMode == 0) ? 0 : 1;
        bestCostIsp[n] = csTemp->cost < bestCostIsp[n] ? csTemp->cost : bestCostIsp[n];
      }

      // check r-d cost
      if (csTemp->cost < csBest->cost)
      {
        validReturn   = true;
        std::swap(csTemp, csBest);
        bestPUMode    = testMode;
        bestLfnstIdx  = csBest->cus[0]->lfnstIdx;
        bestISP       = csBest->cus[0]->ispMode;
        bestMip       = csBest->cus[0]->mipFlag;
        bestMrl       = csBest->cus[0]->multiRefIdx;
        bestbdpcmMode = cu.bdpcmM[CH_L];
        m_ispTestedModes[bestLfnstIdx].bestSplitSoFar = ISPType(bestISP);
        if (csBest->cost < bestCost)
        {
          bestCost = csBest->cost;
        }
        if ((csBest->getTU(partitioner.chType)->mtsIdx[COMP_Y] == MTS_SKIP) && ( floorLog2(csBest->getTU(partitioner.chType)->blocks[COMP_Y].area()) >= 6 ))
        {
          noLFNST_ts = 1;
        }
      }

      // reset context models
      m_CABACEstimator->getCtx() = ctxStart;

      csTemp->releaseIntermediateData();

      if (m_pcEncCfg->m_fastLocalDualTreeMode && CU::isConsIntra(cu) && !cu.slice->isIntra() && csBest->cost != MAX_DOUBLE && costInterCU != COST_UNKNOWN && mode >= 0)
      {
        if( (m_pcEncCfg->m_fastLocalDualTreeMode == 2) || (csBest->cost > costInterCU * 1.5))
        {
          //Note: only try one intra mode, which is especially useful to reduce EncT for LDB case (around 4%)
          EndMode = 0;
          break;
        }
      }
    }
  } // Mode loop

  if (m_pcEncCfg->m_FastIntraTools && (sps.ISP|| sps.LFNST || sps.MTS))
  {
    int bestMode = csBest->getTU(partitioner.chType)->mtsIdx[COMP_Y] ? 4 : 0;
    bestMode |= bestLfnstIdx ? 2 : 0;
    bestMode |= bestISP ? 1 : 0;
    m_ispTestedModes[0].bestIntraMode = bestMode;
  }
  cu.ispMode = bestISP;
  if( validReturn )
  {
    cs.useSubStructure( *csBest, partitioner.chType, TREE_D, cu.singleChan( CH_L ), true );

    //=== update PU data ====
    cu.lfnstIdx           = bestLfnstIdx;
    cu.mipTransposedFlag  = bestPUMode.mipTrFlg;
    cu.intraDir[CH_L]     = bestPUMode.modeId;
    cu.bdpcmM[CH_L]       = bestbdpcmMode;
    cu.mipFlag            = bestMip;
    cu.multiRefIdx        = bestMrl;
  }
  else
  {
    THROW("fix this");
  }

  csBest->releaseIntermediateData();

  return validReturn;
}

void IntraSearch::estIntraPredChromaQT( CodingUnit& cu, Partitioner& partitioner, const double maxCostAllowed )
{
  PROFILER_SCOPE_AND_STAGE_EXT( 0, _TPROF, P_INTRA_CHROMA, cu.cs, CH_C );
  const TempCtx ctxStart( m_CtxCache, m_CABACEstimator->getCtx() );
  CodingStructure &cs   = *cu.cs;
  bool lumaUsesISP      = !CU::isSepTree(cu) && cu.ispMode;
  PartSplit ispType     = lumaUsesISP ? CU::getISPType(cu, COMP_Y) : TU_NO_ISP;
  double bestCostSoFar  = maxCostAllowed;
  const uint32_t numberValidComponents = getNumberValidComponents( cu.chromaFormat );
  const bool useBDPCM   = cs.picture->useBDPCM;

  uint32_t   uiBestMode = 0;
  Distortion uiBestDist = 0;
  double     dBestCost  = MAX_DOUBLE;

  //----- init mode list ----
  {
    uint32_t  uiMinMode = 0;
    uint32_t  uiMaxMode = NUM_CHROMA_MODE;

    const int reducedModeNumber = uiMaxMode >> (m_pcEncCfg->m_reduceIntraChromaModesFullRD ? 1 : 2);
    //----- check chroma modes -----
    uint32_t chromaCandModes[ NUM_CHROMA_MODE ];
    CU::getIntraChromaCandModes( cu, chromaCandModes );

    // create a temporary CS
    CodingStructure &saveCS = *m_pSaveCS[0];
    saveCS.pcv      = cs.pcv;
    saveCS.picture  = cs.picture;
    saveCS.area.repositionTo( cs.area );
    saveCS.clearTUs();

    if( !CU::isSepTree(cu) && cu.ispMode )
    {
      saveCS.clearCUs();
    }

    if( CU::isSepTree(cu) )
    {
      if( partitioner.canSplit( TU_MAX_TR_SPLIT, cs ) )
      {
        partitioner.splitCurrArea( TU_MAX_TR_SPLIT, cs );

        do
        {
          cs.addTU( CS::getArea( cs, partitioner.currArea(), partitioner.chType, partitioner.treeType ), partitioner.chType, &cu ).depth = partitioner.currTrDepth;
        } while( partitioner.nextPart( cs ) );

        partitioner.exitCurrSplit();
      }
      else
        cs.addTU( CS::getArea( cs, partitioner.currArea(), partitioner.chType, partitioner.treeType ), partitioner.chType, &cu );
    }

    // create a store for the TUs
    std::vector<TransformUnit*> orgTUs;
    for( const auto &ptu : cs.tus )
    {
      // for split TUs in HEVC, add the TUs without Chroma parts for correct setting of Cbfs
      if (lumaUsesISP || cu.contains(*ptu, CH_C))
      {
        saveCS.addTU( *ptu, partitioner.chType, nullptr );
        orgTUs.push_back( ptu );
      }
    }

    // SATD pre-selecting.
    int     satdModeList  [NUM_CHROMA_MODE] = { 0 };
    int64_t satdSortedCost[NUM_CHROMA_MODE] = { 0 };
    bool    modeDisable[NUM_INTRA_MODE + 1] = { false }; // use intra mode idx to check whether enable

    CodingStructure& cs = *(cu.cs);
    CompArea areaCb = cu.Cb();
    CompArea areaCr = cu.Cr();
    CPelBuf orgCb  = cs.getOrgBuf (COMP_Cb);
    PelBuf predCb  = cs.getPredBuf(COMP_Cb);
    CPelBuf orgCr  = cs.getOrgBuf (COMP_Cr);
    PelBuf predCr  = cs.getPredBuf(COMP_Cr);

    DistParam distParamSadCb  = m_pcRdCost->setDistParam( orgCb, predCb, cu.cs->sps->bitDepths[ CH_C ], DF_SAD);
    DistParam distParamSatdCb = m_pcRdCost->setDistParam( orgCb, predCb, cu.cs->sps->bitDepths[ CH_C ], DF_HAD);
    DistParam distParamSadCr  = m_pcRdCost->setDistParam( orgCr, predCr, cu.cs->sps->bitDepths[ CH_C ], DF_SAD);
    DistParam distParamSatdCr = m_pcRdCost->setDistParam( orgCr, predCr, cu.cs->sps->bitDepths[ CH_C ], DF_HAD);

    cu.intraDir[1] = MDLM_L_IDX; // temporary assigned, just to indicate this is a MDLM mode. for luma down-sampling operation.

    initIntraPatternChType(cu, cu.Cb());
    initIntraPatternChType(cu, cu.Cr());
    loadLMLumaRecPels(cu, cu.Cb());

    for (int idx = uiMinMode; idx < uiMaxMode; idx++)
    {
      int mode = chromaCandModes[idx];
      satdModeList[idx] = mode;
      if (CU::isLMCMode(mode) && ( !CU::isLMCModeEnabled(cu, mode) || cu.slice->lmChromaCheckDisable ) )
      {
        continue;
      }
      if ((mode == LM_CHROMA_IDX) || (mode == PLANAR_IDX) || (mode == DM_CHROMA_IDX)) // only pre-check regular modes and MDLM modes, not including DM ,Planar, and LM
      {
        continue;
      }

      cu.intraDir[1]    = mode; // temporary assigned, for SATD checking.

      const bool isLMCMode = CU::isLMCMode(mode);
      if( isLMCMode )
      {
        predIntraChromaLM(COMP_Cb, predCb, cu, areaCb, mode);
      }
      else
      {
        initPredIntraParams(cu, cu.Cb(), *cs.sps);
        predIntraAng(COMP_Cb, predCb, cu);
      }
      int64_t sadCb = distParamSadCb.distFunc(distParamSadCb) * 2;
      int64_t satdCb = distParamSatdCb.distFunc(distParamSatdCb);
      int64_t sad = std::min(sadCb, satdCb);

      if( isLMCMode )
      {
        predIntraChromaLM(COMP_Cr, predCr, cu, areaCr, mode);
      }
      else
      {
        initPredIntraParams(cu, cu.Cr(), *cs.sps);
        predIntraAng(COMP_Cr, predCr, cu);
      }
      int64_t sadCr = distParamSadCr.distFunc(distParamSadCr) * 2;
      int64_t satdCr = distParamSatdCr.distFunc(distParamSatdCr);
      sad += std::min(sadCr, satdCr);
      satdSortedCost[idx] = sad;
    }

    // sort the mode based on the cost from small to large.
    for (int i = uiMinMode; i <= uiMaxMode - 1; i++)
    {
      for (int j = i + 1; j <= uiMaxMode - 1; j++)
      {
        if (satdSortedCost[j] < satdSortedCost[i])
        {
          std::swap( satdModeList[i],   satdModeList[j]);
          std::swap( satdSortedCost[i], satdSortedCost[j]);
        }
      }
    }

    for (int i = 0; i < reducedModeNumber; i++)
    {
      modeDisable[satdModeList[uiMaxMode - 1 - i]] = true; // disable the last reducedModeNumber modes
    }

    int bestLfnstIdx = 0;
    // save the dist
    Distortion baseDist = cs.dist;
    int32_t bestbdpcmMode = 0;
    uint32_t numbdpcmModes = ( useBDPCM && CU::bdpcmAllowed(cu, COMP_Cb)
        && ((partitioner.chType == CH_C) || (cu.ispMode == 0 && cu.lfnstIdx == 0 && cu.firstTU->mtsIdx[COMP_Y] == MTS_SKIP))) ? 2 : 0;
    for (int mode_cur = uiMinMode; mode_cur < (int)(uiMaxMode + numbdpcmModes); mode_cur++)
    {
      int mode = mode_cur;
      if (mode_cur >= uiMaxMode)
      {
        mode = mode_cur > uiMaxMode ? -1 : -2; //set bdpcm mode
        if ((mode == -1) && (saveCS.tus[0]->mtsIdx[COMP_Cb] != MTS_SKIP) && (saveCS.tus[0]->mtsIdx[COMP_Cr] != MTS_SKIP))
        {
          continue;
        }
      }
      int chromaIntraMode;
      if (mode < 0)
      {
        cu.bdpcmM[CH_C] = -mode;
        chromaIntraMode = cu.bdpcmM[CH_C] == 2 ? chromaCandModes[1] : chromaCandModes[2];
      }
      else
      {
        cu.bdpcmM[CH_C] = 0;
        chromaIntraMode = chromaCandModes[mode];
        if (CU::isLMCMode(chromaIntraMode) && ( !CU::isLMCModeEnabled(cu, chromaIntraMode) || cu.slice->lmChromaCheckDisable ) )
        {
          continue;
        }
        if (modeDisable[chromaIntraMode] && CU::isLMCModeEnabled(cu, chromaIntraMode)) // when CCLM is disable, then MDLM is disable. not use satd checking
        {
          continue;
        }
      }
      cs.dist = baseDist;
      //----- restore context models -----
      m_CABACEstimator->getCtx() = ctxStart;

      //----- chroma coding -----
      cu.intraDir[1] = chromaIntraMode;
      m_ispTestedModes[0].IspType = ispType;
      m_ispTestedModes[0].subTuCounter = -1;
      xIntraChromaCodingQT( cs, partitioner );
      if (lumaUsesISP && cs.dist == MAX_UINT)
      {
        continue;
      }

      if (cs.sps->transformSkip)
      {
        m_CABACEstimator->getCtx() = ctxStart;
      }
      m_ispTestedModes[0].IspType = ispType;
      m_ispTestedModes[0].subTuCounter = -1;
      uint64_t fracBits   = xGetIntraFracBitsQT( cs, partitioner, false );
      Distortion uiDist = cs.dist;
      double    dCost   = m_pcRdCost->calcRdCost( fracBits, uiDist - baseDist );

      //----- compare -----
      if( dCost < dBestCost )
      {
        if (lumaUsesISP && (dCost < bestCostSoFar))
        {
          bestCostSoFar = dCost;
        }
        for( uint32_t i = getFirstComponentOfChannel( CH_C ); i < numberValidComponents; i++ )
        {
          const CompArea& area = cu.blocks[i];
          saveCS.getRecoBuf     ( area ).copyFrom( cs.getRecoBuf   ( area ) );
          cs.picture->getRecoBuf( area ).copyFrom( cs.getRecoBuf   ( area ) );
          for( uint32_t j = 0; j < saveCS.tus.size(); j++ )
          {
            saveCS.tus[j]->copyComponentFrom( *orgTUs[j], area.compID );
          }
        }
        dBestCost    = dCost;
        uiBestDist   = uiDist;
        uiBestMode   = chromaIntraMode;
        bestLfnstIdx = cu.lfnstIdx;
        bestbdpcmMode = cu.bdpcmM[CH_C];

      }
    }
    cu.lfnstIdx = bestLfnstIdx;
    cu.bdpcmM[CH_C]= bestbdpcmMode;

    for( uint32_t i = getFirstComponentOfChannel( CH_C ); i < numberValidComponents; i++ )
    {
      const CompArea& area = cu.blocks[i];

      cs.getRecoBuf         ( area ).copyFrom( saveCS.getRecoBuf( area ) );
      cs.picture->getRecoBuf( area ).copyFrom( cs.getRecoBuf    ( area ) );

      for( uint32_t j = 0; j < saveCS.tus.size(); j++ )
      {
        orgTUs[ j ]->copyComponentFrom( *saveCS.tus[ j ], area.compID );
      }
    }
  }
  cu.intraDir[1] = uiBestMode;
  cs.dist        = uiBestDist;

  //----- restore context models -----
  m_CABACEstimator->getCtx() = ctxStart;
  if (lumaUsesISP && bestCostSoFar >= maxCostAllowed)
  {
    cu.ispMode = 0;
  }
}

void IntraSearch::saveCuAreaCostInSCIPU( Area area, double cost )
{
  if( m_numCuInSCIPU < NUM_INTER_CU_INFO_SAVE )
  {
    m_cuAreaInSCIPU[m_numCuInSCIPU] = area;
    m_cuCostInSCIPU[m_numCuInSCIPU] = cost;
    m_numCuInSCIPU++;
  }
}

void IntraSearch::initCuAreaCostInSCIPU()
{
  for( int i = 0; i < NUM_INTER_CU_INFO_SAVE; i++ )
  {
    m_cuAreaInSCIPU[i] = Area();
    m_cuCostInSCIPU[i] = 0;
  }
  m_numCuInSCIPU = 0;
}
// -------------------------------------------------------------------------------------------------------------------
// Intra search
// -------------------------------------------------------------------------------------------------------------------

void IntraSearch::xEncIntraHeader( CodingStructure &cs, Partitioner &partitioner, const bool luma )
{
  CodingUnit &cu = *cs.getCU( partitioner.chType, partitioner.treeType );

  if (luma)
  {
    bool isFirst = cu.ispMode ? m_ispTestedModes[0].subTuCounter == 0 : partitioner.currArea().lumaPos() == cs.area.lumaPos();

    // CU header
    if( isFirst )
    {
      if ((!cs.slice->isIntra() || cs.slice->sps->IBC || cs.slice->sps->PLT) && cu.Y().valid())
      {
        m_CABACEstimator->pred_mode   ( cu );
      }
      m_CABACEstimator->bdpcm_mode  ( cu, ComponentID(partitioner.chType) );
    }

    // luma prediction mode
    if (isFirst)
    {
      if ( !cu.Y().valid())
      {
        m_CABACEstimator->pred_mode( cu );
      }
      m_CABACEstimator->intra_luma_pred_mode( cu );
    }
  }
  else //  if (chroma)
  {
    bool isFirst = partitioner.currArea().Cb().valid() && partitioner.currArea().chromaPos() == cs.area.chromaPos();

    if( isFirst )
    {
      m_CABACEstimator->bdpcm_mode(cu, ComponentID(CH_C));
      m_CABACEstimator->intra_chroma_pred_mode(  cu );
    }
  }
}

void IntraSearch::xEncSubdivCbfQT( CodingStructure &cs, Partitioner &partitioner, const bool luma )
{
  const UnitArea& currArea = partitioner.currArea();
  int subTuCounter = m_ispTestedModes[0].subTuCounter;
  TransformUnit  &currTU   = *cs.getTU(currArea.blocks[partitioner.chType], partitioner.chType, subTuCounter);
  CodingUnit     &currCU   = *currTU.cu;
  const uint32_t currDepth = partitioner.currTrDepth;
  const bool  subdiv = currTU.depth > currDepth;
  ComponentID compID = partitioner.chType == CH_L ? COMP_Y : COMP_Cb;

  if (!luma)
  {
    const bool chromaCbfISP = currArea.blocks[COMP_Cb].valid() && currCU.ispMode && !subdiv;
    if (!currCU.ispMode || chromaCbfISP)
    {
      const uint32_t numberValidComponents = getNumberValidComponents(currArea.chromaFormat);
      const uint32_t cbfDepth = (chromaCbfISP ? currDepth - 1 : currDepth);

      for (uint32_t ch = COMP_Cb; ch < numberValidComponents; ch++)
      {
        const ComponentID compID = ComponentID(ch);
        if (currDepth == 0 || TU::getCbfAtDepth(currTU, compID, currDepth - 1) || chromaCbfISP)
        {
          const bool prevCbf = (compID == COMP_Cr ? TU::getCbfAtDepth(currTU, COMP_Cb, currDepth) : false);
          m_CABACEstimator->cbf_comp(currCU, TU::getCbfAtDepth(currTU, compID, currDepth), currArea.blocks[compID], cbfDepth, prevCbf);
        }
      }
    }
  }

  if (subdiv)
  {
    if (partitioner.canSplit(TU_MAX_TR_SPLIT, cs))
    {
      partitioner.splitCurrArea(TU_MAX_TR_SPLIT, cs);
    }
    else if (currCU.ispMode && isLuma(compID))
    {
      partitioner.splitCurrArea(m_ispTestedModes[0].IspType, cs);
    }
    else
      THROW("Cannot perform an implicit split!");

    do
    {
      xEncSubdivCbfQT(cs, partitioner, luma);   //?
      subTuCounter += subTuCounter != -1 ? 1 : 0;
    } while (partitioner.nextPart(cs));

    partitioner.exitCurrSplit();
  }
  else
  {
    //===== Cbfs =====
    if (luma)
    {
      bool previousCbf = false;
      bool lastCbfIsInferred = false;
      if (m_ispTestedModes[0].IspType != TU_NO_ISP)
      {
        bool     rootCbfSoFar = false;
        uint32_t nTus = currCU.ispMode == HOR_INTRA_SUBPARTITIONS ? currCU.lheight() >> floorLog2(currTU.lheight())
          : currCU.lwidth() >> floorLog2(currTU.lwidth());
        if (subTuCounter == nTus - 1)
        {
          TransformUnit* tuPointer = currCU.firstTU;
          for (int tuIdx = 0; tuIdx < nTus - 1; tuIdx++)
          {
            rootCbfSoFar |= TU::getCbfAtDepth(*tuPointer, COMP_Y, currDepth);
            tuPointer = tuPointer->next;
          }
          if (!rootCbfSoFar)
          {
            lastCbfIsInferred = true;
          }
        }
        if (!lastCbfIsInferred)
        {
          previousCbf = TU::getPrevTuCbfAtDepth(currTU, COMP_Y, partitioner.currTrDepth);
        }
      }
      if (!lastCbfIsInferred)
      {
        m_CABACEstimator->cbf_comp(currCU, TU::getCbfAtDepth(currTU, COMP_Y, currDepth), currTU.Y(), currTU.depth, previousCbf, currCU.ispMode);
      }
    }
  }
}
void IntraSearch::xEncCoeffQT(CodingStructure& cs, Partitioner& partitioner, const ComponentID compID, CUCtx* cuCtx, const int subTuIdx, const PartSplit ispType)
{
  const UnitArea& currArea  = partitioner.currArea();

  int subTuCounter          = m_ispTestedModes[0].subTuCounter;
  TransformUnit& currTU     = *cs.getTU(currArea.blocks[partitioner.chType], partitioner.chType, subTuCounter);
  uint32_t   currDepth      = partitioner.currTrDepth;
  const bool subdiv         = currTU.depth > currDepth;

  if (subdiv)
  {
    if (partitioner.canSplit(TU_MAX_TR_SPLIT, cs))
    {
      partitioner.splitCurrArea(TU_MAX_TR_SPLIT, cs);
    }
    else if (currTU.cu->ispMode)
    {
      partitioner.splitCurrArea(m_ispTestedModes[0].IspType, cs);
    }
    else
      THROW("Implicit TU split not available!");

    do
    {
      xEncCoeffQT(cs, partitioner, compID, cuCtx, subTuCounter, m_ispTestedModes[0].IspType);
      subTuCounter += subTuCounter != -1 ? 1 : 0;
    } while( partitioner.nextPart( cs ) );

    partitioner.exitCurrSplit();
  }
  else

  if( currArea.blocks[compID].valid() )
  {
    if( compID == COMP_Cr )
    {
      const int cbfMask = ( TU::getCbf( currTU, COMP_Cb ) ? 2 : 0 ) + ( TU::getCbf( currTU, COMP_Cr ) ? 1 : 0 );
      m_CABACEstimator->joint_cb_cr( currTU, cbfMask );
    }
    if( TU::getCbf( currTU, compID ) )
    {
      if( isLuma(compID) )
      {
        m_CABACEstimator->residual_coding( currTU, compID, cuCtx );
        m_CABACEstimator->mts_idx( *currTU.cu, cuCtx );
      }
      else
        m_CABACEstimator->residual_coding( currTU, compID );
    }
  }
}

uint64_t IntraSearch::xGetIntraFracBitsQT( CodingStructure &cs, Partitioner &partitioner, const bool luma, CUCtx *cuCtx )
{
  m_CABACEstimator->resetBits();

  xEncIntraHeader( cs, partitioner, luma );
  xEncSubdivCbfQT( cs, partitioner, luma );

  if( luma )
  {
    xEncCoeffQT( cs, partitioner, COMP_Y, cuCtx );

    CodingUnit &cu = *cs.cus[0];
    if (cuCtx /*&& CU::isSepTree(cu)*/
      && (!cu.ispMode || (cu.lfnstIdx && m_ispTestedModes[0].subTuCounter == 0)
        || (!cu.lfnstIdx
          && m_ispTestedModes[0].subTuCounter == m_ispTestedModes[cu.lfnstIdx].numTotalParts[cu.ispMode - 1] - 1)))
    {
      m_CABACEstimator->residual_lfnst_mode( cu, *cuCtx );
    }
  }
  else
  {
    xEncCoeffQT( cs, partitioner, COMP_Cb );
    xEncCoeffQT( cs, partitioner, COMP_Cr );
  }

  uint64_t fracBits = m_CABACEstimator->getEstFracBits();
  return fracBits;
}

uint64_t IntraSearch::xGetIntraFracBitsQTChroma(const TransformUnit& currTU, const ComponentID compID, CUCtx *cuCtx)
{
  m_CABACEstimator->resetBits();

  if ( currTU.jointCbCr )
  {
    const int cbfMask = ( TU::getCbf( currTU, COMP_Cb ) ? 2 : 0 ) + ( TU::getCbf( currTU, COMP_Cr ) ? 1 : 0 );
    m_CABACEstimator->cbf_comp( *currTU.cu, cbfMask>>1, currTU.blocks[ COMP_Cb ], currTU.depth, false );
    m_CABACEstimator->cbf_comp( *currTU.cu, cbfMask &1, currTU.blocks[ COMP_Cr ], currTU.depth, cbfMask>>1 );
    if( cbfMask )
      m_CABACEstimator->joint_cb_cr( currTU, cbfMask );
    if (cbfMask >> 1)
      m_CABACEstimator->residual_coding( currTU, COMP_Cb, cuCtx );
    if (cbfMask & 1)
      m_CABACEstimator->residual_coding( currTU, COMP_Cr, cuCtx );
  }
  else
  {
    if ( compID == COMP_Cb )
      m_CABACEstimator->cbf_comp( *currTU.cu, TU::getCbf( currTU, compID ), currTU.blocks[ compID ], currTU.depth, false );
    else
    {
      const bool cbCbf    = TU::getCbf( currTU, COMP_Cb );
      const bool crCbf    = TU::getCbf( currTU, compID );
      const int  cbfMask  = ( cbCbf ? 2 : 0 ) + ( crCbf ? 1 : 0 );
      m_CABACEstimator->cbf_comp( *currTU.cu, crCbf, currTU.blocks[ compID ], currTU.depth, cbCbf );
      m_CABACEstimator->joint_cb_cr( currTU, cbfMask );
    }
  }

  if( !currTU.jointCbCr && TU::getCbf( currTU, compID ) )
  {
    m_CABACEstimator->residual_coding( currTU, compID, cuCtx );
  }

  uint64_t fracBits = m_CABACEstimator->getEstFracBits();
  return fracBits;
}

void IntraSearch::xIntraCodingTUBlock(TransformUnit &tu, const ComponentID compID, const bool checkCrossCPrediction, Distortion &ruiDist, uint32_t *numSig, PelUnitBuf *predBuf, const bool loadTr)
{
  if (!tu.blocks[compID].valid())
  {
    return;
  }

  CodingStructure &cs             = *tu.cs;
  const CompArea      &area       = tu.blocks[compID];
  const SPS           &sps        = *cs.sps;

  const ChannelType    chType     = toChannelType(compID);
  const int            bitDepth   = sps.bitDepths[chType];

  CPelBuf        piOrg            = cs.getOrgBuf    (area);
  PelBuf         piPred           = cs.getPredBuf   (area);
  PelBuf         piResi           = cs.getResiBuf   (area);
  PelBuf         piReco           = cs.getRecoBuf   (area);

  const CodingUnit& cu            = *tu.cu;

  //===== init availability pattern =====
  CHECK( tu.jointCbCr && compID == COMP_Cr, "wrong combination of compID and jointCbCr" );
  bool jointCbCr = tu.jointCbCr && compID == COMP_Cb;

  if ( isLuma(compID) )
  {
    bool predRegDiffFromTB = CU::isPredRegDiffFromTB(*tu.cu );
    bool firstTBInPredReg  = false;
    CompArea areaPredReg(COMP_Y, tu.chromaFormat, area);
    if (tu.cu->ispMode )
    {
      firstTBInPredReg = CU::isFirstTBInPredReg(*tu.cu, area);
      if (predRegDiffFromTB)
      {
        if (firstTBInPredReg)
        {
          CU::adjustPredArea(areaPredReg);
          initIntraPatternChTypeISP(*tu.cu, areaPredReg, piReco);
        }
      }
      else
        initIntraPatternChTypeISP(*tu.cu, area, piReco);
    }
    else if( !predBuf )
    {
      initIntraPatternChType(*tu.cu, area);
    }

    //===== get prediction signal =====
    if (predRegDiffFromTB)
    {
      if (firstTBInPredReg)
      {
        PelBuf piPredReg = cs.getPredBuf(areaPredReg);
        predIntraAng(compID, piPredReg, cu);
      }
    }
    else
    {
      if( predBuf )
      {
        piPred.copyFrom( predBuf->Y() );
      }
      else if( CU::isMIP( cu, CH_L ) )
      {
        initIntraMip( cu );
        predIntraMip( piPred, cu );
      }
      else
      {
        predIntraAng(compID, piPred, cu);
      }
    }
  }
  DTRACE( g_trace_ctx, D_PRED, "@(%4d,%4d) [%2dx%2d] IMode=%d\n", tu.lx(), tu.ly(), tu.lwidth(), tu.lheight(), CU::getFinalIntraMode(cu, chType) );

  if (isLuma(compID))
  {
    //===== get residual signal =====
    piResi.subtract( piOrg, piPred );
  }

  //===== transform and quantization =====
  //--- init rate estimation arrays for RDOQ ---
  //--- transform and quantization           ---
  TCoeff uiAbsSum = 0;
  const QpParam cQP(tu, compID);

  m_pcTrQuant->selectLambda(compID);

  if ( jointCbCr )
  {
    // Lambda is loosened for the joint mode with respect to single modes as the same residual is used for both chroma blocks
    const int    absIct = abs( TU::getICTMode(tu) );
    const double lfact  = ( absIct == 1 || absIct == 3 ? 0.8 : 0.5 );
    m_pcTrQuant->scaleLambda( lfact );
  }
  if ( sps.jointCbCr && isChroma(compID) && (tu.cu->cs->slice->sliceQp > 18) )
  {
    m_pcTrQuant->scaleLambda( 1.3 );
  }

  if( isLuma(compID) )
  {
    m_pcTrQuant->transformNxN(tu, compID, cQP, uiAbsSum, m_CABACEstimator->getCtx(), loadTr);

    DTRACE( g_trace_ctx, D_TU_ABS_SUM, "%d: comp=%d, abssum=%d\n", DTRACE_GET_COUNTER( g_trace_ctx, D_TU_ABS_SUM ), compID, uiAbsSum );
    if (tu.cu->ispMode && isLuma(compID) && CU::isISPLast(*tu.cu, area, area.compID) && CU::allLumaCBFsAreZero(*tu.cu))
    {
      // ISP has to have at least one non-zero CBF
      ruiDist = MAX_INT;
      return;
    }
    //--- inverse transform ---
    if (uiAbsSum > 0)
    {
      m_pcTrQuant->invTransformNxN(tu, compID, piResi, cQP);
    }
    else
    {
      piResi.fill(0);
    }
  }
  else // chroma
  {
    PelBuf          crPred = cs.getPredBuf ( COMP_Cr );
    PelBuf          crResi = cs.getResiBuf ( COMP_Cr );
    PelBuf          crReco = cs.getRecoBuf ( COMP_Cr );

    int         codedCbfMask  = 0;
    ComponentID codeCompId    = (tu.jointCbCr ? (tu.jointCbCr >> 1 ? COMP_Cb : COMP_Cr) : compID);
    const QpParam qpCbCr(tu, codeCompId);

    if( tu.jointCbCr )
    {
      ComponentID otherCompId = ( codeCompId==COMP_Cr ? COMP_Cb : COMP_Cr );
      tu.getCoeffs( otherCompId ).fill(0); // do we need that?
      TU::setCbfAtDepth (tu, otherCompId, tu.depth, false );
    }
    PelBuf& codeResi = ( codeCompId == COMP_Cr ? crResi : piResi );
    uiAbsSum = 0;
    m_pcTrQuant->transformNxN(tu, codeCompId, qpCbCr, uiAbsSum, m_CABACEstimator->getCtx(), loadTr);
    DTRACE( g_trace_ctx, D_TU_ABS_SUM, "%d: comp=%d, abssum=%d\n", DTRACE_GET_COUNTER( g_trace_ctx, D_TU_ABS_SUM ), codeCompId, uiAbsSum );
    if( uiAbsSum > 0 )
    {
      m_pcTrQuant->invTransformNxN(tu, codeCompId, codeResi, qpCbCr);
      codedCbfMask += ( codeCompId == COMP_Cb ? 2 : 1 );
    }
    else
    {
      codeResi.fill(0);
    }

    if( tu.jointCbCr )
    {
      if( tu.jointCbCr == 3 && codedCbfMask == 2 )
      {
        codedCbfMask = 3;
        TU::setCbfAtDepth (tu, COMP_Cr, tu.depth, true );
      }
      if( tu.jointCbCr != codedCbfMask )
      {
        ruiDist = MAX_DISTORTION;
        return;
      }
      m_pcTrQuant->invTransformICT( tu, piResi, crResi );
      uiAbsSum = codedCbfMask;
    }

    //===== reconstruction =====
    if( jointCbCr )
    {
      crReco.reconstruct(crPred, crResi, cs.slice->clpRngs[ COMP_Cr ]);
    }
  }
  piReco.reconstruct(piPred, piResi, cs.slice->clpRngs[ compID ]);
  


  //===== update distortion =====
  ruiDist += m_pcRdCost->getDistPart( piOrg, piReco, bitDepth, compID, DF_SSE );
  if( jointCbCr )
  {
    CPelBuf         crOrg  = cs.getOrgBuf  ( COMP_Cr );
    PelBuf          crReco = cs.getRecoBuf ( COMP_Cr );
    ruiDist += m_pcRdCost->getDistPart( crOrg, crReco, bitDepth, COMP_Cr, DF_SSE );
  }
}

void IntraSearch::xIntraCodingLumaQT(CodingStructure& cs, Partitioner& partitioner, PelUnitBuf* predBuf, const double bestCostSoFar, int numMode, bool disableMTS)
{
  PROFILER_SCOPE_AND_STAGE_EXT( 0, _TPROF, P_INTRA_RD_SEARCH_LUMA, &cs, partitioner.chType );
  const UnitArea& currArea  = partitioner.currArea();
  uint32_t        currDepth = partitioner.currTrDepth;
  Distortion singleDistLuma = 0;
  uint32_t   numSig         = 0;
  const SPS &sps            = *cs.sps;
  CodingUnit &cu            = *cs.cus[0];
  bool mtsAllowed = (numMode < 0) || disableMTS ? false : CU::isMTSAllowed(cu, COMP_Y);
  uint64_t singleFracBits   = 0;
  bool   splitCbfLumaSum    = false;
  double bestCostForISP     = bestCostSoFar;
  double dSingleCost        = MAX_DOUBLE;
  int endLfnstIdx           = (partitioner.isSepTree(cs) && partitioner.chType == CH_C && (currArea.lwidth() < 8 || currArea.lheight() < 8))
                           || (currArea.lwidth() > sps.getMaxTbSize() || currArea.lheight() > sps.getMaxTbSize()) || !sps.LFNST || (numMode < 0) ? 0 : 2;
  const bool useTS          = cs.picture->useTS;
  numMode                   = (numMode < 0) ? -numMode : numMode;

  if (cu.mipFlag && !allowLfnstWithMip(cu.lumaSize()))
  {
    endLfnstIdx = 0;
  }
  int bestMTS = 0;
  int EndMTS  = mtsAllowed ? m_pcEncCfg->m_MTSIntraMaxCand : 0;
  if (cu.ispMode && (EndMTS || endLfnstIdx))
  {
    EndMTS = 0;
    if ((m_ispTestedModes[1].numTotalParts[cu.ispMode - 1] == 0)
     && (m_ispTestedModes[2].numTotalParts[cu.ispMode - 1] == 0))
    {
      endLfnstIdx = 0;
    }
  }
  if (cu.bdpcmM[CH_L])
  {
    endLfnstIdx = 0;
    EndMTS = 0;
  }
  bool checkTransformSkip = sps.transformSkip;

  SizeType transformSkipMaxSize = 1 << sps.log2MaxTransformSkipBlockSize;
  bool tsAllowed = useTS  && cu.cs->sps->transformSkip && (!cu.ispMode) && (!cu.bdpcmM[CH_L]) && (!cu.sbtInfo);
  tsAllowed &= cu.blocks[COMP_Y].width <= transformSkipMaxSize && cu.blocks[COMP_Y].height <= transformSkipMaxSize;
  if (tsAllowed)
  {
    EndMTS += 1;
  }
  if (endLfnstIdx || EndMTS)
  {
    bool       splitCbfLuma  = false;
    const PartSplit ispType  = CU::getISPType(cu, COMP_Y);
    CUCtx cuCtx;
    cuCtx.isDQPCoded         = true;
    cuCtx.isChromaQpAdjCoded = true;
    cs.cost                  = 0.0;
    Distortion       singleDistTmpLuma = 0;
    uint64_t         singleTmpFracBits = 0;
    double           singleCostTmp     = 0;
    const TempCtx    ctxStart          (m_CtxCache, m_CABACEstimator->getCtx());
          TempCtx    ctxBest           (m_CtxCache);
    CodingStructure &saveCS            = *m_pSaveCS[cu.ispMode?0:1];
    TransformUnit *  tmpTU             = nullptr;
    int              bestLfnstIdx      = 0;
    int              startLfnstIdx     = 0;
    // speedUps LFNST
    bool   rapidLFNST                  = false;
    bool   rapidDCT                    = false;
    double thresholdDCT                = 1;

    if (m_pcEncCfg->m_MTS == 2)
    {
      thresholdDCT += 1.4 / sqrt(cu.lwidth() * cu.lheight());
    }

    if (m_pcEncCfg->m_LFNST > 1)
    {
      rapidLFNST = true;

      if (m_pcEncCfg->m_LFNST > 2)
      {
        rapidDCT    = true;
        endLfnstIdx = endLfnstIdx ? 1 : 0;
      }
    }

    saveCS.pcv              = cs.pcv;
    saveCS.picture          = cs.picture;
    saveCS.area.repositionTo( cs.area);

    if (cu.ispMode)
    {
      partitioner.splitCurrArea(ispType, cs);
    }

    TransformUnit& tu = cs.addTU(CS::getArea(cs, partitioner.currArea(), partitioner.chType, partitioner.treeType), partitioner.chType, cs.cus[0]);

    if (cu.ispMode)
    {
      saveCS.clearTUs();
      do
      {
        saveCS.addTU(
          CS::getArea(cs, partitioner.currArea(), partitioner.chType, partitioner.treeType),
          partitioner.chType, cs.cus[0]);
      } while (partitioner.nextPart(cs));

      partitioner.exitCurrSplit();
    }
    else
    {
      tmpTU = saveCS.tus.empty() ? &saveCS.addTU( currArea, partitioner.chType, nullptr ) : saveCS.tus.front();
      tmpTU->initData();
      tmpTU->UnitArea::operator=( currArea );
    }


    std::vector<TrMode> trModes{ TrMode(0, true) };
    if (tsAllowed)
    {
      trModes.push_back(TrMode(1, true));
    }
    double dct2Cost           = MAX_DOUBLE;
    double trGrpStopThreshold = 1.001;
    double trGrpBestCost      = MAX_DOUBLE;

    if (mtsAllowed)
    {
      if (m_pcEncCfg->m_LFNST)
      {
        uint32_t uiIntraMode = cs.cus[0]->intraDir[partitioner.chType];
        int MTScur           = (uiIntraMode < 34) ? MTS_DST7_DCT8 : MTS_DCT8_DST7;

        trModes.push_back(TrMode(     2, true));
        trModes.push_back(TrMode(MTScur, true));

        MTScur = (uiIntraMode < 34) ? MTS_DCT8_DST7 : MTS_DST7_DCT8;

        trModes.push_back(TrMode(MTScur,            true));
        trModes.push_back(TrMode(MTS_DST7_DST7 + 3, true));
      }
      else
      {
        for (int i = 2; i < 6; i++)
        {
          trModes.push_back(TrMode(i, true));
        }
      }
    }

    if ((EndMTS && !m_pcEncCfg->m_LFNST) || (tsAllowed && !mtsAllowed))
    {
      xPreCheckMTS(tu, &trModes, m_pcEncCfg->m_MTSIntraMaxCand, predBuf);
      if (!mtsAllowed && !trModes[1].second)
      {
        EndMTS = 0;
      }
    }

    bool NStopMTS = true;

    for (int modeId = 0; modeId <= EndMTS && NStopMTS; modeId++)
    {
      if (modeId > 1)
      {
        trGrpBestCost = MAX_DOUBLE;
      }
      for (int lfnstIdx = startLfnstIdx; lfnstIdx <= endLfnstIdx; lfnstIdx++)
      {
        if (lfnstIdx && modeId)
        {
          continue;
        }
        if (mtsAllowed || tsAllowed)
        {
          if (m_pcEncCfg->m_TS && bestMTS == MTS_SKIP)
          {
            break;
          }
          if (!m_pcEncCfg->m_LFNST && !trModes[modeId].second && mtsAllowed)
          {
            continue;
          }

          tu.mtsIdx[COMP_Y] = trModes[modeId].first;
        }

        if (cu.ispMode && lfnstIdx)
        {
          if (m_ispTestedModes[lfnstIdx].numTotalParts[cu.ispMode - 1] == 0)
          {
            if (lfnstIdx == 2)
            {
              endLfnstIdx = 1;
            }
            continue;
          }
        }

        cu.lfnstIdx                          = lfnstIdx;
        cuCtx.lfnstLastScanPos               = false;
        cuCtx.violatesLfnstConstrained[CH_L] = false;
        cuCtx.violatesLfnstConstrained[CH_C] = false;

        if ((lfnstIdx != startLfnstIdx) || (modeId))
        {
          m_CABACEstimator->getCtx() = ctxStart;
        }

        singleDistTmpLuma = 0;

        if (cu.ispMode)
        {
          splitCbfLuma = false;

          partitioner.splitCurrArea(ispType, cs);

          singleCostTmp = xTestISP(cs, partitioner, bestCostForISP, ispType, splitCbfLuma, singleTmpFracBits, singleDistTmpLuma, cuCtx);

          partitioner.exitCurrSplit();

          if (modeId && (singleCostTmp == MAX_DOUBLE))
          {
            m_ispTestedModes[lfnstIdx].numTotalParts[cu.ispMode - 1] = 0;
          }

          bool storeCost = (numMode == 1) ? true : false;

          if ((m_pcEncCfg->m_ISP >= 2) && (numMode <= 1))
          {
            storeCost = true;
          }

          if (storeCost)
          {
            m_ispTestedModes[0].bestCost[cu.ispMode - 1] = singleCostTmp;
          }
        }
        else
        {
          bool TrLoad = (EndMTS && !m_pcEncCfg->m_LFNST) || (tsAllowed && !mtsAllowed && (lfnstIdx == 0)) ? true : false;

          xIntraCodingTUBlock(tu, COMP_Y, false, singleDistTmpLuma, &numSig, predBuf, TrLoad);

          cuCtx.mtsLastScanPos = false;
          //----- determine rate and r-d cost -----
        if ((sps.LFNST ? (modeId == EndMTS && modeId != 0 && checkTransformSkip) : (trModes[modeId].first != 0)) && !TU::getCbfAtDepth(tu, COMP_Y, currDepth))
        {
          singleCostTmp = MAX_DOUBLE;
        }
        else
        {
          m_ispTestedModes[0].IspType      = TU_NO_ISP;
          m_ispTestedModes[0].subTuCounter = -1;
          singleTmpFracBits = xGetIntraFracBitsQT(cs, partitioner, true, &cuCtx);

          if (tu.mtsIdx[COMP_Y] > MTS_SKIP)
          {
            if (!cuCtx.mtsLastScanPos)
            {
              singleCostTmp = MAX_DOUBLE;
            }
            else
            {
              singleCostTmp = m_pcRdCost->calcRdCost(singleTmpFracBits, singleDistTmpLuma);
            }
          }
          else
          {
            singleCostTmp = m_pcRdCost->calcRdCost(singleTmpFracBits, singleDistTmpLuma);
          }
        }

          if (((EndMTS && (m_pcEncCfg->m_MTS == 2)) || rapidLFNST) && modeId == 0 && lfnstIdx == 0)
          {
            if (singleCostTmp > bestCostSoFar * thresholdDCT)
            {
              EndMTS = 0;

              if (rapidDCT)
              {
                endLfnstIdx = 0;   // break the loop but do not cpy best
              }
            }
          }

          if (lfnstIdx && !cuCtx.lfnstLastScanPos && !cu.ispMode)
          {
            bool rootCbfL = false;

            for (uint32_t t = 0; t < getNumberValidTBlocks(*cu.cs->pcv); t++)
            {
              rootCbfL |= tu.cbf[t] != 0;
            }

            if (rapidLFNST && !rootCbfL)
            {
              endLfnstIdx = lfnstIdx; // break the loop
            }
            bool cbfAtZeroDepth = CU::isSepTree(cu)
              ? rootCbfL
              : (cs.area.chromaFormat != CHROMA_400 && std::min(cu.firstTU->blocks[1].width, cu.firstTU->blocks[1].height) < 4)
                ? TU::getCbfAtDepth(tu, COMP_Y, currDepth)
                : rootCbfL;

            if (cbfAtZeroDepth)
            {
              singleCostTmp = MAX_DOUBLE;
            }
          }
        }

        if (singleCostTmp < dSingleCost)
        {
          trGrpBestCost  = singleCostTmp;
          dSingleCost    = singleCostTmp;
          singleDistLuma = singleDistTmpLuma;
          singleFracBits = singleTmpFracBits;
          bestLfnstIdx   = lfnstIdx;
          bestMTS        = modeId;

          if (dSingleCost < bestCostForISP)
          {
            bestCostForISP = dSingleCost;
          }

          splitCbfLumaSum = splitCbfLuma;

          if (lfnstIdx == 0 && modeId == 0 && cu.ispMode == 0)
          {
            dct2Cost = singleCostTmp;

            if (!TU::getCbfAtDepth(tu, COMP_Y, currDepth))
            {
              if (rapidLFNST)
              {
                 endLfnstIdx = 0;   // break the loop but do not cpy best
              }

              EndMTS = 0;
            }
          }

          if (bestLfnstIdx != endLfnstIdx || bestMTS != EndMTS)
          {
            if (cu.ispMode)
            {
              saveCS.getRecoBuf(currArea.Y()).copyFrom(cs.getRecoBuf(currArea.Y()));

              for (uint32_t j = 0; j < cs.tus.size(); j++)
              {
                saveCS.tus[j]->copyComponentFrom(*cs.tus[j], COMP_Y);
              }
            }
            else
            {
              saveCS.getPredBuf(tu.Y()).copyFrom(cs.getPredBuf(tu.Y()));
              saveCS.getRecoBuf(tu.Y()).copyFrom(cs.getRecoBuf(tu.Y()));

              tmpTU->copyComponentFrom(tu, COMP_Y);
            }

            ctxBest = m_CABACEstimator->getCtx();
          }
      
        }
        else
        {
          if( rapidLFNST )
          {
            endLfnstIdx = lfnstIdx; // break the loop
          }
        }
      }
      if (m_pcEncCfg->m_LFNST && m_pcEncCfg->m_MTS == 2 && modeId && modeId != EndMTS)
      {
        NStopMTS = false;

        if (bestMTS || bestLfnstIdx)
        {
          if ((modeId > 1 && bestMTS == modeId) || modeId == 1)
          {
            NStopMTS = (dct2Cost / trGrpBestCost) < trGrpStopThreshold;
          }
        }
      }
    }

    cu.lfnstIdx = bestLfnstIdx;
    if (dSingleCost != MAX_DOUBLE)
    {
      if (bestLfnstIdx != endLfnstIdx || bestMTS != EndMTS)
      {
        if (cu.ispMode)
        {
          const UnitArea& currArea = partitioner.currArea();
          cs.getRecoBuf(currArea.Y()).copyFrom(saveCS.getRecoBuf(currArea.Y()));

          if (saveCS.tus.size() != cs.tus.size())
          {
            partitioner.splitCurrArea(ispType, cs);

            do
            {
              partitioner.nextPart(cs);
              cs.addTU(CS::getArea(cs, partitioner.currArea(), partitioner.chType, partitioner.treeType),
                partitioner.chType, cs.cus[0]);
            } while (saveCS.tus.size() != cs.tus.size());

            partitioner.exitCurrSplit();
          }

          for (uint32_t j = 0; j < saveCS.tus.size(); j++)
          {
            cs.tus[j]->copyComponentFrom(*saveCS.tus[j], COMP_Y);
          }
        }
        else
        {
          cs.getRecoBuf(tu.Y()).copyFrom(saveCS.getRecoBuf(tu.Y()));

          tu.copyComponentFrom(*tmpTU, COMP_Y);
        }

        m_CABACEstimator->getCtx() = ctxBest;
      }

      // otherwise this would've happened in useSubStructure
      cs.picture->getRecoBuf(currArea.Y()).copyFrom(cs.getRecoBuf(currArea.Y()));
    }
  }
  else
  {
    if (cu.ispMode)
    {
      const PartSplit ispType = CU::getISPType(cu, COMP_Y);
      partitioner.splitCurrArea(ispType, cs);

      CUCtx      cuCtx;
      dSingleCost = xTestISP(cs, partitioner, bestCostForISP, ispType, splitCbfLumaSum, singleFracBits, singleDistLuma, cuCtx);
      partitioner.exitCurrSplit();
      bool storeCost = (numMode == 1) ? true : false;
      if ((m_pcEncCfg->m_ISP >= 2) && (numMode <= 1))
      {
        storeCost = true;
      }
      if (storeCost)
      {
        m_ispTestedModes[0].bestCost[cu.ispMode - 1] = dSingleCost;
      }
    }
    else
    {
      TransformUnit& tu =
        cs.addTU(CS::getArea(cs, currArea, partitioner.chType, partitioner.treeType), partitioner.chType, cs.cus[0]);
      tu.depth = currDepth;

      CHECK(!tu.Y().valid(), "Invalid TU");
      xIntraCodingTUBlock(tu, COMP_Y, false, singleDistLuma, &numSig, predBuf);
      //----- determine rate and r-d cost -----
      m_ispTestedModes[0].IspType = TU_NO_ISP;
      m_ispTestedModes[0].subTuCounter = -1;
      singleFracBits = xGetIntraFracBitsQT(cs, partitioner, true);
      dSingleCost = m_pcRdCost->calcRdCost(singleFracBits, singleDistLuma);
    }
  }

  if (cu.ispMode)
  { 
    for (auto& ptu : cs.tus)
    {
      if (currArea.Y().contains(ptu->Y()))
      {
        TU::setCbfAtDepth(*ptu, COMP_Y, currDepth, splitCbfLumaSum ? 1 : 0);
      }
    }
  }
  cs.dist     += singleDistLuma;
  cs.fracBits += singleFracBits;
  cs.cost      = dSingleCost;

  STAT_COUNT_CU_MODES( partitioner.chType == CH_L, g_cuCounters1D[CU_RD_TESTS][0][!cs.slice->isIntra() + cs.slice->depth] );
  STAT_COUNT_CU_MODES( partitioner.chType == CH_L && !cs.slice->isIntra(), g_cuCounters2D[CU_RD_TESTS][Log2( cs.area.lheight() )][Log2( cs.area.lwidth() )] );
}

ChromaCbfs IntraSearch::xIntraChromaCodingQT(CodingStructure& cs, Partitioner& partitioner)
{
  UnitArea    currArea      = partitioner.currArea();

  if( !currArea.Cb().valid() ) 
    return ChromaCbfs(false);

  TransformUnit& currTU     = *cs.getTU( currArea.chromaPos(), CH_C );
  const CodingUnit& cu  = *cs.getCU( currArea.chromaPos(), CH_C, TREE_D );
  ChromaCbfs cbfs(false);
  uint32_t   currDepth = partitioner.currTrDepth;
  const bool useTS = cs.picture->useTS;
  if (currDepth == currTU.depth)
  {
    if (!currArea.Cb().valid() || !currArea.Cr().valid())
    {
      return cbfs;
    }

    CodingStructure& saveCS = *m_pSaveCS[1];
    saveCS.pcv = cs.pcv;
    saveCS.picture = cs.picture;
    saveCS.area.repositionTo(cs.area);

    TransformUnit& tmpTU = saveCS.tus.empty() ? saveCS.addTU(currArea, partitioner.chType, nullptr) : *saveCS.tus.front();
    tmpTU.initData();
    tmpTU.UnitArea::operator=(currArea);
    const unsigned      numTBlocks = getNumberValidTBlocks(*cs.pcv);

    CompArea& cbArea = currTU.blocks[COMP_Cb];
    CompArea& crArea = currTU.blocks[COMP_Cr];
    double     bestCostCb = MAX_DOUBLE;
    double     bestCostCr = MAX_DOUBLE;
    Distortion bestDistCb = 0;
    Distortion bestDistCr = 0;

    TempCtx ctxStartTU(m_CtxCache);
    TempCtx ctxStart(m_CtxCache);
    TempCtx ctxBest(m_CtxCache);

    ctxStartTU = m_CABACEstimator->getCtx();
    ctxStart = m_CABACEstimator->getCtx();
    currTU.jointCbCr = 0;

    // Do predictions here to avoid repeating the "default0Save1Load2" stuff
    int  predMode = cu.bdpcmM[CH_C] ? BDPCM_IDX : CU::getFinalIntraMode(cu, CH_C);

    PelBuf piPredCb = cs.getPredBuf(COMP_Cb);
    PelBuf piPredCr = cs.getPredBuf(COMP_Cr);

    initIntraPatternChType(*currTU.cu, cbArea);
    initIntraPatternChType(*currTU.cu, crArea);

    if (CU::isLMCMode(predMode))
    {
      loadLMLumaRecPels(cu, cbArea);
      predIntraChromaLM(COMP_Cb, piPredCb, cu, cbArea, predMode);
      predIntraChromaLM(COMP_Cr, piPredCr, cu, crArea, predMode);
    }
    else
    {
      predIntraAng(COMP_Cb, piPredCb, cu);
      predIntraAng(COMP_Cr, piPredCr, cu);
    }

    // determination of chroma residuals including reshaping and cross-component prediction
    //----- get chroma residuals -----
    PelBuf resiCb = cs.getResiBuf(COMP_Cb);
    PelBuf resiCr = cs.getResiBuf(COMP_Cr);
    resiCb.subtract(cs.getOrgBuf(COMP_Cb), piPredCb);
    resiCr.subtract(cs.getOrgBuf(COMP_Cr), piPredCr);

    //===== store original residual signals (std and crossCompPred) =====
    for( int k = 0; k < 5; k++ )
    {
      m_orgResiCb[k].compactResize( cbArea );
      m_orgResiCr[k].compactResize( crArea );
    }
    for (int k = 0; k < 1; k += 4)
    {
      m_orgResiCb[k].copyFrom(resiCb);
      m_orgResiCr[k].copyFrom(resiCr);
    }

    CUCtx cuCtx;
    cuCtx.isDQPCoded = true;
    cuCtx.isChromaQpAdjCoded = true;
    cuCtx.lfnstLastScanPos = false;

    CodingStructure& saveCScur = *m_pSaveCS[2];

    saveCScur.pcv = cs.pcv;
    saveCScur.picture = cs.picture;
    saveCScur.area.repositionTo(cs.area);

    TransformUnit& tmpTUcur = saveCScur.tus.empty() ? saveCScur.addTU(currArea, partitioner.chType, nullptr) : *saveCScur.tus.front();
    tmpTUcur.initData();
    tmpTUcur.UnitArea::operator=(currArea);

    TempCtx ctxBestTUL(m_CtxCache);

    const SPS& sps = *cs.sps;
    double     bestCostCbcur = MAX_DOUBLE;
    double     bestCostCrcur = MAX_DOUBLE;
    Distortion bestDistCbcur = 0;
    Distortion bestDistCrcur = 0;

    int  endLfnstIdx = (partitioner.isSepTree(cs) && partitioner.chType == CH_C && (partitioner.currArea().lwidth() < 8 || partitioner.currArea().lheight() < 8))
      || (partitioner.currArea().lwidth() > sps.getMaxTbSize() || partitioner.currArea().lheight() > sps.getMaxTbSize()) || !sps.LFNST ? 0 : 2;
    int  startLfnstIdx = 0;
    int  bestLfnstIdx = 0;
    bool testLFNST = sps.LFNST;

    // speedUps LFNST
    bool rapidLFNST = false;
    if (m_pcEncCfg->m_LFNST > 1)
    {
      rapidLFNST = true;
      if (m_pcEncCfg->m_LFNST > 2)
      {
        endLfnstIdx = endLfnstIdx ? 1 : 0;
      }
    }
    int ts_used = 0;
    bool testTS = false;
    if (partitioner.chType != CH_C)
    {
      startLfnstIdx = currTU.cu->lfnstIdx;
      endLfnstIdx = currTU.cu->lfnstIdx;
      bestLfnstIdx = currTU.cu->lfnstIdx;
      testLFNST  = false;
      rapidLFNST = false;
      ts_used = currTU.mtsIdx[COMP_Y];
    }
    if (cu.bdpcmM[CH_C])
    {
      endLfnstIdx = 0;
      testLFNST = false;
    }

    double dSingleCostAll = MAX_DOUBLE;
    double singleCostTmpAll = 0;

    for (int lfnstIdx = startLfnstIdx; lfnstIdx <= endLfnstIdx; lfnstIdx++)
    {
      if (rapidLFNST && lfnstIdx)
      {
        if ((lfnstIdx == 2) && (bestLfnstIdx == 0))
        {
          continue;
        }
      }

      currTU.cu->lfnstIdx = lfnstIdx;
      if (lfnstIdx)
      {
        m_CABACEstimator->getCtx() = ctxStartTU;
      }

      cuCtx.lfnstLastScanPos = false;
      cuCtx.violatesLfnstConstrained[CH_L] = false;
      cuCtx.violatesLfnstConstrained[CH_C] = false;

      for (uint32_t c = COMP_Cb; c < numTBlocks; c++)
      {
        const ComponentID compID = ComponentID(c);
        const CompArea& area = currTU.blocks[compID];
        double     dSingleCost = MAX_DOUBLE;
        Distortion singleDistCTmp = 0;
        double     singleCostTmp = 0;
        bool tsAllowed = useTS && TU::isTSAllowed(currTU, compID) && m_pcEncCfg->m_useChromaTS && !currTU.cu->lfnstIdx && !cu.bdpcmM[CH_C];
        if ((partitioner.chType == CH_L) && (!ts_used))
        {
          tsAllowed = false;
        }
        uint8_t nNumTransformCands = 1 + (tsAllowed ? 1 : 0); // DCT + TS = 2 tests       
        std::vector<TrMode> trModes;
        if (nNumTransformCands > 1)
        {
          trModes.push_back(TrMode(0, true));   // DCT2
          trModes.push_back(TrMode(1, true));   // TS
          testTS = true;
        }
        bool cbfDCT2 = true;
        const bool isLastMode = testLFNST || cs.sps->jointCbCr ||  tsAllowed ? false : true;
        int bestModeId = 0;
        ctxStart = m_CABACEstimator->getCtx();
        for (int modeId = 0; modeId < nNumTransformCands; modeId++)
        {
          if (lfnstIdx || modeId)
          {
            resiCb.copyFrom(m_orgResiCb[0]);
            resiCr.copyFrom(m_orgResiCr[0]);
          }
          if (modeId == 0)
          {
            if ( tsAllowed)
            {
              xPreCheckMTS(currTU, &trModes, m_pcEncCfg->m_MTSIntraMaxCand, 0, compID);
            }
          }

          currTU.mtsIdx[compID] = currTU.cu->bdpcmM[CH_C] ? MTS_SKIP : modeId;

          if (modeId)
          {
            if (!cbfDCT2 && trModes[modeId].first == MTS_SKIP)
            {
              break;
            }
            m_CABACEstimator->getCtx() = ctxStart;
          }
          singleDistCTmp = 0;
          if (tsAllowed)
          {
            xIntraCodingTUBlock(currTU, compID, false, singleDistCTmp, 0, 0, true);
            if ((modeId == 0) && (!trModes[modeId + 1].second))
            {
              nNumTransformCands = 1;
            }
          }
          else
        {
          xIntraCodingTUBlock(currTU, compID, false, singleDistCTmp);
        }
        if (((currTU.mtsIdx[compID] == MTS_SKIP && !currTU.cu->bdpcmM[CH_C])
          && !TU::getCbf(currTU, compID)))   // In order not to code TS flag when cbf is zero, the case for TS with
                                             // cbf being zero is forbidden.
        {
          singleCostTmp = MAX_DOUBLE;
        }
        else
        {
          uint64_t fracBitsTmp = xGetIntraFracBitsQTChroma(currTU, compID, &cuCtx);
          singleCostTmp = m_pcRdCost->calcRdCost(fracBitsTmp, singleDistCTmp);
        }

        if (singleCostTmp < dSingleCost)
        {
          dSingleCost = singleCostTmp;

          if (compID == COMP_Cb)
          {
            bestCostCb = singleCostTmp;
            bestDistCb = singleDistCTmp;
          }
          else
          {
            bestCostCr = singleCostTmp;
            bestDistCr = singleDistCTmp;
          }
          bestModeId = modeId;
          if (currTU.mtsIdx[compID] == MTS_DCT2_DCT2)
          {
            cbfDCT2 = TU::getCbfAtDepth(currTU, compID, currDepth);
          }
          if (!isLastMode)
          {
            saveCS.getRecoBuf(area).copyFrom(cs.getRecoBuf(area));
            tmpTU.copyComponentFrom(currTU, compID);
            ctxBest = m_CABACEstimator->getCtx();
          }
        }
        }
        if (testTS && ((c == COMP_Cb && bestModeId < (nNumTransformCands - 1)) ))
        {
          m_CABACEstimator->getCtx() = ctxBest;

          currTU.copyComponentFrom(tmpTU, COMP_Cb); // Cbf of Cb is needed to estimate cost for Cr Cbf
        }
      }

      singleCostTmpAll = bestCostCb + bestCostCr;

      bool rootCbfL = false;
      if (testLFNST)
      {
        for (uint32_t t = 0; t < getNumberValidTBlocks(*cs.pcv); t++)
        {
          rootCbfL |= bool(tmpTU.cbf[t]);
        }
        if (rapidLFNST && !rootCbfL)
        {
          endLfnstIdx = lfnstIdx; // end this
        }
      }

      if (testLFNST && lfnstIdx && !cuCtx.lfnstLastScanPos)
      {
        bool cbfAtZeroDepth = CU::isSepTree(*currTU.cu)
          ? rootCbfL : (cs.area.chromaFormat != CHROMA_400
            && std::min(tmpTU.blocks[1].width, tmpTU.blocks[1].height) < 4)
          ? TU::getCbfAtDepth(currTU, COMP_Y, currTU.depth) : rootCbfL;
        if (cbfAtZeroDepth)
        {
          singleCostTmpAll = MAX_DOUBLE;
        }
      }
      if ((testLFNST || testTS) && (singleCostTmpAll < dSingleCostAll))
      {
        bestLfnstIdx = lfnstIdx;
        if ((lfnstIdx != endLfnstIdx) || testTS)
        {
          dSingleCostAll = singleCostTmpAll;

          bestCostCbcur = bestCostCb;
          bestCostCrcur = bestCostCr;
          bestDistCbcur = bestDistCb;
          bestDistCrcur = bestDistCr;

          saveCScur.getRecoBuf(cbArea).copyFrom(saveCS.getRecoBuf(cbArea));
          saveCScur.getRecoBuf(crArea).copyFrom(saveCS.getRecoBuf(crArea));

          tmpTUcur.copyComponentFrom(tmpTU, COMP_Cb);
          tmpTUcur.copyComponentFrom(tmpTU, COMP_Cr);
        }
        ctxBestTUL = m_CABACEstimator->getCtx();
      }
    }
    if ((testLFNST && (bestLfnstIdx != endLfnstIdx)) || testTS)
    {
      bestCostCb = bestCostCbcur;
      bestCostCr = bestCostCrcur;
      bestDistCb = bestDistCbcur;
      bestDistCr = bestDistCrcur;
      currTU.cu->lfnstIdx = bestLfnstIdx;
      if (!cs.sps->jointCbCr)
      {
        cs.getRecoBuf(cbArea).copyFrom(saveCScur.getRecoBuf(cbArea));
        cs.getRecoBuf(crArea).copyFrom(saveCScur.getRecoBuf(crArea));

        currTU.copyComponentFrom(tmpTUcur, COMP_Cb);
        currTU.copyComponentFrom(tmpTUcur, COMP_Cr);

        m_CABACEstimator->getCtx() = ctxBestTUL;
      }
    }

    Distortion bestDistCbCr = bestDistCb + bestDistCr;

    if (cs.sps->jointCbCr)
    {
      if ((testLFNST && (bestLfnstIdx != endLfnstIdx)) || testTS)
      {
        saveCS.getRecoBuf(cbArea).copyFrom(saveCScur.getRecoBuf(cbArea));
        saveCS.getRecoBuf(crArea).copyFrom(saveCScur.getRecoBuf(crArea));

        tmpTU.copyComponentFrom(tmpTUcur, COMP_Cb);
        tmpTU.copyComponentFrom(tmpTUcur, COMP_Cr);
        m_CABACEstimator->getCtx() = ctxBestTUL;
        ctxBest = m_CABACEstimator->getCtx();
      }
      // Test using joint chroma residual coding
      double     bestCostCbCr = bestCostCb + bestCostCr;
      int        bestJointCbCr = 0;
      bool checkDCTOnly = m_pcEncCfg->m_useChromaTS && ((TU::getCbf(tmpTU, COMP_Cb) && tmpTU.mtsIdx[COMP_Cb] == MTS_DCT2_DCT2 && !TU::getCbf(tmpTU, COMP_Cr)) ||
        (TU::getCbf(tmpTU, COMP_Cr) && tmpTU.mtsIdx[COMP_Cr] == MTS_DCT2_DCT2 && !TU::getCbf(tmpTU, COMP_Cb)) ||
        (TU::getCbf(tmpTU, COMP_Cb) && tmpTU.mtsIdx[COMP_Cb] == MTS_DCT2_DCT2 && TU::getCbf(tmpTU, COMP_Cr) && tmpTU.mtsIdx[COMP_Cr] == MTS_DCT2_DCT2));
      bool checkTSOnly = m_pcEncCfg->m_useChromaTS && ((TU::getCbf(tmpTU, COMP_Cb) && tmpTU.mtsIdx[COMP_Cb] == MTS_SKIP && !TU::getCbf(tmpTU, COMP_Cr)) ||
        (TU::getCbf(tmpTU, COMP_Cr) && tmpTU.mtsIdx[COMP_Cr] == MTS_SKIP && !TU::getCbf(tmpTU, COMP_Cb)) ||
        (TU::getCbf(tmpTU, COMP_Cb) && tmpTU.mtsIdx[COMP_Cb] == MTS_SKIP && TU::getCbf(tmpTU, COMP_Cr) && tmpTU.mtsIdx[COMP_Cr] == MTS_SKIP));
      bool       lastIsBest = false;
      bool noLFNST1 = false;
      if (rapidLFNST && (startLfnstIdx != endLfnstIdx))
      {
        if (bestLfnstIdx == 2)
        {
          noLFNST1 = true;
        }
        else
        {
          endLfnstIdx = 1;
        }
      }

      for (int lfnstIdxj = startLfnstIdx; lfnstIdxj <= endLfnstIdx; lfnstIdxj++)
      {
        if (rapidLFNST && noLFNST1 && (lfnstIdxj == 1))
        {
          continue;
        }
        currTU.cu->lfnstIdx = lfnstIdxj;
        std::vector<int> jointCbfMasksToTest;
        if (TU::getCbf(tmpTU, COMP_Cb) || TU::getCbf(tmpTU, COMP_Cr))
        {
          jointCbfMasksToTest = m_pcTrQuant->selectICTCandidates(currTU, m_orgResiCb, m_orgResiCr);
        }
        for (int cbfMask : jointCbfMasksToTest)
        {
          currTU.jointCbCr = (uint8_t)cbfMask;
          ComponentID codeCompId = ((currTU.jointCbCr >> 1) ? COMP_Cb : COMP_Cr);
          ComponentID otherCompId = ((codeCompId == COMP_Cb) ? COMP_Cr : COMP_Cb);
          bool tsAllowed = useTS && TU::isTSAllowed(currTU, codeCompId) && (m_pcEncCfg->m_useChromaTS) && !currTU.cu->lfnstIdx && !cu.bdpcmM[CH_C];
          if ((partitioner.chType == CH_L)&& tsAllowed && (currTU.mtsIdx[COMP_Y] != MTS_SKIP))
          {
            tsAllowed = false;
          }
          if (!tsAllowed)
          {
            checkTSOnly = false;
          }
          uint8_t     numTransformCands = 1 + (tsAllowed && !(checkDCTOnly || checkTSOnly)? 1 : 0); // DCT + TS = 2 tests
          std::vector<TrMode> trModes;
          if (numTransformCands > 1)
          {
            trModes.push_back(TrMode(0, true)); // DCT2
            trModes.push_back(TrMode(1, true));//TS
          }
          else
          {
            currTU.mtsIdx[codeCompId] = checkTSOnly || currTU.cu->bdpcmM[CH_C] ? 1 : 0;
          }

          for (int modeId = 0; modeId < numTransformCands; modeId++)
          {
            Distortion distTmp = 0;
            currTU.mtsIdx[codeCompId] = currTU.cu->bdpcmM[CH_C] ? MTS_SKIP : MTS_DCT2_DCT2;
            if (numTransformCands > 1)
            {
              currTU.mtsIdx[codeCompId] = currTU.cu->bdpcmM[CH_C] ? MTS_SKIP : trModes[modeId].first;
            }
            currTU.mtsIdx[otherCompId] = MTS_DCT2_DCT2;

            m_CABACEstimator->getCtx() = ctxStartTU;

            resiCb.copyFrom(m_orgResiCb[cbfMask]);
            resiCr.copyFrom(m_orgResiCr[cbfMask]);
            if ((modeId == 0) && (numTransformCands > 1))
            {
              xPreCheckMTS(currTU, &trModes, m_pcEncCfg->m_MTSIntraMaxCand, 0, COMP_Cb);
              currTU.mtsIdx[codeCompId] = trModes[modeId].first;
              currTU.mtsIdx[(codeCompId == COMP_Cr) ? COMP_Cb : COMP_Cr] = MTS_DCT2_DCT2;
            }
            cuCtx.lfnstLastScanPos = false;
            cuCtx.violatesLfnstConstrained[CH_L] = false;
            cuCtx.violatesLfnstConstrained[CH_C] = false;
            if (numTransformCands > 1)
            {
              xIntraCodingTUBlock(currTU, COMP_Cb, false, distTmp, 0, 0, true);
              if ((modeId == 0) && !trModes[modeId + 1].second)
              {
                numTransformCands = 1;
              }
            }
            else
            {
              xIntraCodingTUBlock(currTU, COMP_Cb, false, distTmp, 0);
            }

            double costTmp = std::numeric_limits<double>::max();
            if (distTmp < MAX_DISTORTION)
            {
              uint64_t bits = xGetIntraFracBitsQTChroma(currTU, COMP_Cb, &cuCtx);
              costTmp = m_pcRdCost->calcRdCost(bits, distTmp);
            }
            else if (!currTU.mtsIdx[codeCompId])
            {
              numTransformCands = 1;
            }
            bool rootCbfL = false;
            for (uint32_t t = 0; t < getNumberValidTBlocks(*cs.pcv); t++)
            {
              rootCbfL |= bool(tmpTU.cbf[t]);
            }
            if (rapidLFNST && !rootCbfL)
            {
              endLfnstIdx = lfnstIdxj;
            }
            if (testLFNST && currTU.cu->lfnstIdx && !cuCtx.lfnstLastScanPos)
            {
              bool cbfAtZeroDepth = CU::isSepTree(*currTU.cu) ? rootCbfL
                : (cs.area.chromaFormat != CHROMA_400 && std::min(tmpTU.blocks[1].width, tmpTU.blocks[1].height) < 4)
                ? TU::getCbfAtDepth(currTU, COMP_Y, currTU.depth) : rootCbfL;
              if (cbfAtZeroDepth)
              {
                costTmp = MAX_DOUBLE;
              }
            }
            if (costTmp < bestCostCbCr)
            {
              bestCostCbCr = costTmp;
              bestDistCbCr = distTmp;
              bestJointCbCr = currTU.jointCbCr;

              // store data
              bestLfnstIdx = lfnstIdxj;
              if ((cbfMask != jointCbfMasksToTest.back() || (lfnstIdxj != endLfnstIdx)) || (modeId != (numTransformCands - 1)))
              {
                saveCS.getRecoBuf(cbArea).copyFrom(cs.getRecoBuf(cbArea));
                saveCS.getRecoBuf(crArea).copyFrom(cs.getRecoBuf(crArea));

                tmpTU.copyComponentFrom(currTU, COMP_Cb);
                tmpTU.copyComponentFrom(currTU, COMP_Cr);

                ctxBest = m_CABACEstimator->getCtx();
              }
              else
              {
                lastIsBest = true;
                cs.cus[0]->lfnstIdx = bestLfnstIdx;
              }
            }
          }
        }

        // Retrieve the best CU data (unless it was the very last one tested)
      }
      if (!lastIsBest)
      {
        cs.getRecoBuf(cbArea).copyFrom(saveCS.getRecoBuf(cbArea));
        cs.getRecoBuf(crArea).copyFrom(saveCS.getRecoBuf(crArea));

        cs.cus[0]->lfnstIdx = bestLfnstIdx;
        currTU.copyComponentFrom(tmpTU, COMP_Cb);
        currTU.copyComponentFrom(tmpTU, COMP_Cr);
        m_CABACEstimator->getCtx() = ctxBest;
      }
      currTU.jointCbCr = (TU::getCbf(currTU, COMP_Cb) || TU::getCbf(currTU, COMP_Cr)) ? bestJointCbCr : 0;
    } // jointCbCr

    cs.dist += bestDistCbCr;
    cuCtx.violatesLfnstConstrained[CH_L] = false;
    cuCtx.violatesLfnstConstrained[CH_C] = false;
    cuCtx.lfnstLastScanPos = false;
    cuCtx.violatesMtsCoeffConstraint = false;
    cuCtx.mtsLastScanPos = false;
    cbfs.cbf(COMP_Cb) = TU::getCbf(currTU, COMP_Cb);
    cbfs.cbf(COMP_Cr) = TU::getCbf(currTU, COMP_Cr);
  }
  else
  {
    unsigned   numValidTBlocks = getNumberValidTBlocks(*cs.pcv);
    ChromaCbfs SplitCbfs(false);

    if (partitioner.canSplit(TU_MAX_TR_SPLIT, cs))
    {
      partitioner.splitCurrArea(TU_MAX_TR_SPLIT, cs);
    }
    else if (currTU.cu->ispMode)
    {
      partitioner.splitCurrArea(m_ispTestedModes[0].IspType, cs);
    }
    else
      THROW("Implicit TU split not available");

    do
    {
      ChromaCbfs subCbfs = xIntraChromaCodingQT(cs, partitioner);

      for (uint32_t ch = COMP_Cb; ch < numValidTBlocks; ch++)
      {
        const ComponentID compID = ComponentID(ch);
        SplitCbfs.cbf(compID) |= subCbfs.cbf(compID);
      }
    } while (partitioner.nextPart(cs));

    partitioner.exitCurrSplit();

    /*if (lumaUsesISP && cs.dist == MAX_UINT) //ahenkel
    {
      return cbfs;
    }*/
    {
      cbfs.Cb |= SplitCbfs.Cb;
      cbfs.Cr |= SplitCbfs.Cr;

      if (1)   //(!lumaUsesISP)
      {
        for (auto& ptu : cs.tus)
        {
          if (currArea.Cb().contains(ptu->Cb()) || (!ptu->Cb().valid() && currArea.Y().contains(ptu->Y())))
          {
            TU::setCbfAtDepth(*ptu, COMP_Cb, currDepth, SplitCbfs.Cb);
            TU::setCbfAtDepth(*ptu, COMP_Cr, currDepth, SplitCbfs.Cr);
          }
        }
      }
    }
  }
  return cbfs;
}

uint64_t IntraSearch::xFracModeBitsIntraLuma(const CodingUnit& cu, const unsigned* mpmLst)
{
  m_CABACEstimator->resetBits();

  if (!cu.ciip)
  {
    m_CABACEstimator->intra_luma_pred_mode(cu, mpmLst);
  }

  return m_CABACEstimator->getEstFracBits();
}

template<typename T, size_t N, int M>
void IntraSearch::xReduceHadCandList(static_vector<T, N>& candModeList, static_vector<double, N>& candCostList, SortedPelUnitBufs<M>& sortedPelBuffer, int& numModesForFullRD, const double thresholdHadCost, const double* mipHadCost, const CodingUnit& cu, const bool fastMip)
{
  const int maxCandPerType = numModesForFullRD >> 1;
  static_vector<ModeInfo, FAST_UDI_MAX_RDMODE_NUM> tempRdModeList;
  static_vector<double, FAST_UDI_MAX_RDMODE_NUM> tempCandCostList;
  const double minCost = candCostList[0];
  bool keepOneMip = candModeList.size() > numModesForFullRD;
  const int maxNumConv = 3; 

  int numConv = 0;
  int numMip = 0;
  for (int idx = 0; idx < candModeList.size() - (keepOneMip?0:1); idx++)
  {
    bool addMode = false;
    const ModeInfo& orgMode = candModeList[idx];

    if (!orgMode.mipFlg)
    {
      addMode = (numConv < maxNumConv);
      numConv += addMode ? 1:0;
    }
    else
    {
      addMode = ( numMip < maxCandPerType || (candCostList[idx] < thresholdHadCost * minCost) || keepOneMip );
      keepOneMip = false;
      numMip += addMode ? 1:0;
    }
    if( addMode )
    {
      tempRdModeList.push_back(orgMode);
      tempCandCostList.push_back(candCostList[idx]);
    }
  }

  // sort Pel Buffer
  int i = -1;
  for( auto &m: tempRdModeList)
  {
    if( ! (m == candModeList.at( ++i )) )
    {
      for( int j = i; j < (int)candModeList.size()-1; )
      {
        if( m == candModeList.at( ++j ) )
        {
          sortedPelBuffer.swap( i, j);
          break;
        }
      }
    }
  }
  sortedPelBuffer.reduceTo( (int)tempRdModeList.size() );

  if ((cu.lwidth() > 8 && cu.lheight() > 8))
  {
    // Sort MIP candidates by Hadamard cost
    const int transpOff = getNumModesMip(cu.Y());
    static_vector<uint8_t, FAST_UDI_MAX_RDMODE_NUM> sortedMipModes(0);
    static_vector<double, FAST_UDI_MAX_RDMODE_NUM> sortedMipCost(0);
    for (uint8_t mode : { 0, 1, 2 })
    {
      uint8_t candMode = mode + uint8_t((mipHadCost[mode + transpOff] < mipHadCost[mode]) ? transpOff : 0);
      updateCandList(candMode, mipHadCost[candMode], sortedMipModes, sortedMipCost, 3);
    }

    // Append MIP mode to RD mode list
    const int modeListSize = int(tempRdModeList.size());
    for (int idx = 0; idx < 3; idx++)
    {
      const bool     isTransposed = (sortedMipModes[idx] >= transpOff ? true : false);
      const uint32_t mipIdx       = (isTransposed ? sortedMipModes[idx] - transpOff : sortedMipModes[idx]);
      const ModeInfo mipMode( true, isTransposed, 0, NOT_INTRA_SUBPARTITIONS, mipIdx );
      bool alreadyIncluded = false;
      for (int modeListIdx = 0; modeListIdx < modeListSize; modeListIdx++)
      {
        if (tempRdModeList[modeListIdx] == mipMode)
        {
          alreadyIncluded = true;
          break;
        }
      }

      if (!alreadyIncluded)
      {
        tempRdModeList.push_back(mipMode);
        tempCandCostList.push_back(0);
        if( fastMip ) break;
      }
    }
  }

  candModeList = tempRdModeList;
  candCostList = tempCandCostList;
  numModesForFullRD = int(candModeList.size());
}

void IntraSearch::xPreCheckMTS(TransformUnit &tu, std::vector<TrMode> *trModes, const int maxCand, PelUnitBuf *predBuf, const ComponentID& compID)
{
  if (compID == COMP_Y)
  {
    CodingStructure&  cs = *tu.cs;
    const CompArea& area = tu.blocks[compID];
    const CodingUnit& cu = *cs.getCU(area.pos(), CH_L,TREE_D);
    PelBuf piPred = cs.getPredBuf(area);
    PelBuf piResi = cs.getResiBuf(area);

    initIntraPatternChType(*tu.cu, area);
    if (predBuf)
    {
      piPred.copyFrom(predBuf->Y());
    }
    else if (CU::isMIP(cu, CH_L))
    {
      initIntraMip(cu);
      predIntraMip(piPred, cu);
    }
    else
    {
      predIntraAng(COMP_Y, piPred, cu);
    }

    //===== get residual signal =====
    CPelBuf piOrg = cs.getOrgBuf(COMP_Y);
    piResi.subtract(piOrg, piPred);
    m_pcTrQuant->checktransformsNxN(tu, trModes, m_pcEncCfg->m_MTSIntraMaxCand, compID);
  }
  else
  {
    ComponentID codeCompId = (tu.jointCbCr ? (tu.jointCbCr >> 1 ? COMP_Cb : COMP_Cr) : compID);
    m_pcTrQuant->checktransformsNxN(tu, trModes, m_pcEncCfg->m_MTSIntraMaxCand, codeCompId);
  }
}

double IntraSearch::xTestISP(CodingStructure& cs, Partitioner& subTuPartitioner, double bestCostForISP, PartSplit ispType, bool& splitcbf, uint64_t& singleFracBits, Distortion& singleDistLuma, CUCtx& cuCtx)
{
  int  subTuCounter = 0;
  bool earlySkipISP = false;
  bool splitCbfLuma = false;
  CodingUnit& cu = *cs.cus[0];

  Distortion singleDistTmpLumaSUM = 0;
  uint64_t   singleTmpFracBitsSUM = 0;
  double     singleCostTmpSUM = 0;
  cuCtx.isDQPCoded = true;
  cuCtx.isChromaQpAdjCoded = true;

  do
  {
    Distortion singleDistTmpLuma = 0;
    uint64_t   singleTmpFracBits = 0;
    double     singleCostTmp = 0;
    TransformUnit& tmpTUcur = ((cs.tus.size() < (subTuCounter + 1)))
      ? cs.addTU(CS::getArea(cs, subTuPartitioner.currArea(), subTuPartitioner.chType,
        subTuPartitioner.treeType),
        subTuPartitioner.chType, cs.cus[0])
      : *cs.tus[subTuCounter];
    tmpTUcur.depth = subTuPartitioner.currTrDepth;

    // Encode TU
    xIntraCodingTUBlock(tmpTUcur, COMP_Y, false, singleDistTmpLuma, 0);
    cuCtx.mtsLastScanPos = false;

    if (singleDistTmpLuma == MAX_INT)   // all zero CBF skip
    {
      earlySkipISP = true;
      singleCostTmpSUM = MAX_DOUBLE;
      break;
    }

    if (m_pcRdCost->calcRdCost(singleTmpFracBitsSUM, singleDistTmpLumaSUM + singleDistTmpLuma) > bestCostForISP)
    {
      earlySkipISP = true;
    }
    else
    {
      m_ispTestedModes[0].IspType = ispType;
      m_ispTestedModes[0].subTuCounter = subTuCounter;
      singleTmpFracBits = xGetIntraFracBitsQT(cs, subTuPartitioner, true, &cuCtx);
    }
    singleCostTmp = m_pcRdCost->calcRdCost(singleTmpFracBits, singleDistTmpLuma);

    singleCostTmpSUM     += singleCostTmp;
    singleDistTmpLumaSUM += singleDistTmpLuma;
    singleTmpFracBitsSUM += singleTmpFracBits;

    subTuCounter++;

    splitCbfLuma |= TU::getCbfAtDepth( *cs.getTU(subTuPartitioner.currArea().lumaPos(), subTuPartitioner.chType, subTuCounter - 1), 
                                       COMP_Y, subTuPartitioner.currTrDepth);
    int nSubPartitions = m_ispTestedModes[cu.lfnstIdx].numTotalParts[cu.ispMode - 1];
    bool doStop = (m_pcEncCfg->m_ISP != 1) || (subTuCounter < nSubPartitions);
    if (doStop)
    {
      if (singleCostTmpSUM > bestCostForISP)
      {
        earlySkipISP = true;
        break;
      }
      if (subTuCounter < nSubPartitions)
      {
        double threshold = nSubPartitions == 2 ? 0.95 : subTuCounter == 1 ? 0.83 : 0.91;
        if (singleCostTmpSUM > bestCostForISP * threshold)
        {
          earlySkipISP = true;
          break;
        }
      }
    }
  } while (subTuPartitioner.nextPart(cs));
  singleDistLuma = singleDistTmpLumaSUM;
  singleFracBits = singleTmpFracBitsSUM;

  splitcbf = splitCbfLuma;
  return earlySkipISP ? MAX_DOUBLE : singleCostTmpSUM;
}

int IntraSearch::xSpeedUpISP(int speed, bool& testISP, int mode, int& noISP, int& endISP, CodingUnit& cu, static_vector<ModeInfo, FAST_UDI_MAX_RDMODE_NUM>& RdModeList, const ModeInfo& bestPUMode, int bestISP, int bestLfnstIdx)
{
  if (speed)
  {
    if (mode >= 1)
    {
      if (m_ispTestedModes[0].splitIsFinished[1] && m_ispTestedModes[0].splitIsFinished[0])
      {
        testISP = false;
        endISP = 0;
      }
      else
      {
        if (m_pcEncCfg->m_ISP >= 2)
        {
          if (mode == 1) //best Hor||Ver
          {
            int bestDir = 0;
            for (int d = 0; d < 2; d++)
            {
              int d2 = d ? 0 : 1;
              if ((m_ispTestedModes[0].bestCost[d] <= m_ispTestedModes[0].bestCost[d2])
                && (m_ispTestedModes[0].bestCost[d] != MAX_DOUBLE))
              {
                bestDir = d + 1;
                m_ispTestedModes[0].splitIsFinished[d2] = true;
              }
            }
            m_ispTestedModes[0].bestModeSoFar = bestDir;
            if (m_ispTestedModes[0].bestModeSoFar <= 0)
            {
              m_ispTestedModes[0].splitIsFinished[1] = true;
              m_ispTestedModes[0].splitIsFinished[0] = true;
              testISP = false;
              endISP = 0;
            }
          }
          if (m_ispTestedModes[0].bestModeSoFar == 2)
          {
            noISP = 1;
          }
          else
          {
            endISP = 1;
          }
        }
      }
    }
    if (testISP)
    {
      if (mode == 2)
      {
        for (int d = 0; d < 2; d++)
        {
          int d2 = d ? 0 : 1;
          if (m_ispTestedModes[0].bestCost[d] == MAX_DOUBLE)
          {
            m_ispTestedModes[0].splitIsFinished[d] = true;
          }
          if ((m_ispTestedModes[0].bestCost[d2] < 1.3 * m_ispTestedModes[0].bestCost[d])
            && (int(m_ispTestedModes[0].bestSplitSoFar) != (d + 1)))
          {
            if (d)
            {
              endISP = 1;
            }
            else
            {
              noISP = 1;
            }
            m_ispTestedModes[0].splitIsFinished[d] = true;
          }
        }
      }
      else
      {
        if (m_ispTestedModes[0].splitIsFinished[0])
        {
          noISP = 1;
        }
        if (m_ispTestedModes[0].splitIsFinished[1])
        {
          endISP = 1;
        }
      }
    }
    if ((noISP == 1) && (endISP == 1))
    {
      endISP = 0;
    }
  }
  else
  {
    bool stopFound = false;
    if (m_pcEncCfg->m_ISP >= 3)
    {
      if (mode)
      {
        if ((bestISP == 0) || ((bestPUMode.modeId != RdModeList[mode - 1].modeId)
          && (bestPUMode.modeId != RdModeList[mode].modeId)))
        {
          stopFound = true;
        }
      }
    }
    if (cu.mipFlag || cu.multiRefIdx)
    {
      cu.mipFlag = false;
      cu.multiRefIdx = 0;
      if (!stopFound)
      {
        for (int k = 0; k < mode; k++)
        {
          if (cu.intraDir[CH_L] == RdModeList[k].modeId)
          {
            stopFound = true;
            break;
          }
        }
      }
    }
    if (stopFound)
    {
      testISP = false;
      endISP = 0;
      return 1;
    }
    if (!stopFound && (m_pcEncCfg->m_ISP >= 2) && (cu.intraDir[CH_L] == DC_IDX))
    {
      stopFound = true;
      endISP = 0;
      return 1;
    }
  }
  return 0;
}

void IntraSearch::xSpeedUpIntra(double bestcost, int& EndMode, int& speedIntra, CodingUnit& cu)
{
  int bestIdxbefore = m_ispTestedModes[0].bestIntraMode;
  if (m_ispTestedModes[0].isIntra)
  {
    if (bestIdxbefore == 1)//ISP
    {
      speedIntra = 14;
    }
    if (bestIdxbefore == 4)//MTS
    {
      speedIntra = 3;
    }
  }
  else if (!cu.cs->slice->isIntra())
  {
    if (bestcost != MAX_DOUBLE)
    {
      speedIntra = 10;
    }
  }
  if (m_ispTestedModes[0].bestBefore[0] == -1)
  {
    speedIntra |= 7;
    if (m_pcEncCfg->m_FastIntraTools == 2)
    {
      EndMode = 1;
    }
  }
  if (!cu.cs->slice->isIntra())
  {
    if ((m_ispTestedModes[0].bestBefore[1] == 1) || (m_ispTestedModes[0].bestBefore[2] == 1))
    {
      speedIntra |= 2;
    }
    if ((m_ispTestedModes[0].bestBefore[1] == 4) || (m_ispTestedModes[0].bestBefore[2] == 4))
    {
      speedIntra |= 3;
    }
    if ((m_ispTestedModes[0].bestBefore[1] == 2) || (m_ispTestedModes[0].bestBefore[2] == 2))
    {
      speedIntra |= 1;
    }
  }
}

} // namespace vvenc

//! \}


Coverage Report

Created: 2026-09-02 06:43