EncGOP.cpp

    {
      pcSlice->setDefaultClpRng( *pcSlice->getSPS() );
    }

    // Allocate some coders, now the number of tiles are known.
    const uint32_t numberOfCtusInFrame = pcPic->cs->pcv->sizeInCtus;
    const int numSubstreamsColumns = (pcSlice->getPPS()->getNumTileColumnsMinus1() + 1);
    const int numSubstreamRows     = pcSlice->getPPS()->getEntropyCodingSyncEnabledFlag() ? pcPic->cs->pcv->heightInCtus : (pcSlice->getPPS()->getNumTileRowsMinus1() + 1);
    const int numSubstreams        = std::max<int> (numSubstreamRows * numSubstreamsColumns, (int) pcPic->brickMap->bricks.size());
    std::vector<OutputBitstream> substreamsOut(numSubstreams);

#if ENABLE_QPA
    pcPic->m_uEnerHpCtu.resize (numberOfCtusInFrame);
    pcPic->m_iOffsetCtu.resize (numberOfCtusInFrame);
#if ENABLE_QPA_SUB_CTU
    if (pcSlice->getPPS()->getUseDQP() && pcSlice->getPPS()->getCuQpDeltaSubdiv() > 0)
    {
      const PreCalcValues &pcv = *pcPic->cs->pcv;
#if MAX_TB_SIZE_SIGNALLING
      const unsigned   mtsLog2 = (unsigned)g_aucLog2[std::min (pcPic->cs->sps->getMaxTbSize(), pcv.maxCUWidth)];
#else
      const unsigned   mtsLog2 = (unsigned)g_aucLog2[std::min<uint32_t> (MAX_TB_SIZEY, pcv.maxCUWidth)];
#endif
      pcPic->m_subCtuQP.resize ((pcv.maxCUWidth >> mtsLog2) * (pcv.maxCUHeight >> mtsLog2));
    }
#endif
#endif
    if (pcSlice->getSPS()->getSAOEnabledFlag())
    {
      pcPic->resizeSAO( numberOfCtusInFrame, 0 );
      pcPic->resizeSAO( numberOfCtusInFrame, 1 );
    }

    // it is used for signalling during CTU mode decision, i.e. before ALF processing
    if( pcSlice->getSPS()->getALFEnabledFlag() )
    {
      pcPic->resizeAlfCtuEnableFlag( numberOfCtusInFrame );
      pcPic->resizeAlfCtbFilterIndex(numberOfCtusInFrame);
    }

    bool decPic = false;
    bool encPic = false;
    // test if we can skip the picture entirely or decode instead of encoding
    trySkipOrDecodePicture( decPic, encPic, *m_pcCfg, pcPic );

    pcPic->cs->slice = pcSlice; // please keep this
#if ENABLE_QPA
    if (pcSlice->getPPS()->getSliceChromaQpFlag() && CS::isDualITree (*pcSlice->getPic()->cs) && !m_pcCfg->getUsePerceptQPA() && (m_pcCfg->getSliceChromaOffsetQpPeriodicity() == 0))
#else
    if (pcSlice->getPPS()->getSliceChromaQpFlag() && CS::isDualITree (*pcSlice->getPic()->cs))
#endif
    {
      // overwrite chroma qp offset for dual tree
      pcSlice->setSliceChromaQpDelta(COMPONENT_Cb, m_pcCfg->getChromaCbQpOffsetDualTree());
      pcSlice->setSliceChromaQpDelta(COMPONENT_Cr, m_pcCfg->getChromaCrQpOffsetDualTree());
      pcSlice->setSliceChromaQpDelta(JOINT_CbCr,   m_pcCfg->getChromaCbCrQpOffsetDualTree());
      m_pcSliceEncoder->setUpLambda(pcSlice, pcSlice->getLambdas()[0], pcSlice->getSliceQp());
    }
    if (pcSlice->getSPS()->getUseReshaper())
    {
      m_pcReshaper->getReshapeCW()->rspTid = pcSlice->getTLayer() + (pcSlice->isIntra() ? 0 : 1);
      m_pcReshaper->getReshapeCW()->rspSliceQP = pcSlice->getSliceQp();

      m_pcReshaper->setSrcReshaped(false);
      m_pcReshaper->setRecReshaped(true);

      if (m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_PQ)
      {
        m_pcReshaper->preAnalyzerHDR(pcPic, pcSlice->getSliceType(), m_pcCfg->getReshapeCW(), m_pcCfg->getDualITree());
      }
      else if (m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_SDR)
      {
        m_pcReshaper->preAnalyzerSDR(pcPic, pcSlice->getSliceType(), m_pcCfg->getReshapeCW(), m_pcCfg->getDualITree());
      }
      else
      {
        THROW("Reshaper for other signal currently not defined!");
      }

      if (pcSlice->getSliceType() == I_SLICE )
      {
        if (m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_PQ)
        {
          m_pcReshaper->initLUTfromdQPModel();
          m_pcEncLib->getRdCost()->updateReshapeLumaLevelToWeightTableChromaMD(m_pcReshaper->getInvLUT());
        }
        else if (m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_SDR)
        {
          if (m_pcReshaper->getReshapeFlag())
          {
            m_pcReshaper->constructReshaperSDR();
            m_pcEncLib->getRdCost()->updateReshapeLumaLevelToWeightTable(m_pcReshaper->getSliceReshaperInfo(), m_pcReshaper->getWeightTable(), m_pcReshaper->getCWeight());
          }
        }
        else
        {
          THROW("Reshaper for other signal currently not defined!");
        }

        m_pcReshaper->setCTUFlag(false);

        //reshape original signal
        if (m_pcReshaper->getSliceReshaperInfo().getUseSliceReshaper())
        {
          pcPic->getOrigBuf(COMPONENT_Y).rspSignal(m_pcReshaper->getFwdLUT());
          m_pcReshaper->setSrcReshaped(true);
          m_pcReshaper->setRecReshaped(true);
        }
      }
      else
      {
        if (!m_pcReshaper->getReshapeFlag())
        {
          m_pcReshaper->setCTUFlag(false);
        }
        else
          m_pcReshaper->setCTUFlag(true);

        m_pcReshaper->getSliceReshaperInfo().setSliceReshapeModelPresentFlag(false);

        if (m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_PQ)
        {
          m_pcEncLib->getRdCost()->restoreReshapeLumaLevelToWeightTable();
        }
        else if (m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_SDR)
        {
          int modIP = pcPic->getPOC() - pcPic->getPOC() / m_pcCfg->getReshapeCW().rspFpsToIp * m_pcCfg->getReshapeCW().rspFpsToIp;
          if (m_pcReshaper->getReshapeFlag() && m_pcCfg->getReshapeCW().rspIntraPeriod == -1 && modIP == 0)           // for LDB, update reshaping curve every second
          {
            m_pcReshaper->getSliceReshaperInfo().setSliceReshapeModelPresentFlag(true);
            m_pcReshaper->constructReshaperSDR();
            m_pcEncLib->getRdCost()->updateReshapeLumaLevelToWeightTable(m_pcReshaper->getSliceReshaperInfo(), m_pcReshaper->getWeightTable(), m_pcReshaper->getCWeight());
          }
        }
        else
        {
          THROW("Reshaper for other signal currently not defined!");
        }
      }

      //set all necessary information in LMCS APS and slice
      pcSlice->setLmcsEnabledFlag(m_pcReshaper->getSliceReshaperInfo().getUseSliceReshaper());
      pcSlice->setLmcsChromaResidualScaleFlag(m_pcReshaper->getSliceReshaperInfo().getSliceReshapeChromaAdj() == 1);
      if (m_pcReshaper->getSliceReshaperInfo().getSliceReshapeModelPresentFlag())
      {
        int apsId = 0;
        pcSlice->setLmcsAPSId(apsId);
        APS* lmcsAPS = pcSlice->getLmcsAPS();
        if (lmcsAPS == nullptr)
        {
          ParameterSetMap<APS> *apsMap = m_pcEncLib->getApsMap();
          lmcsAPS = apsMap->getPS((apsId << NUM_APS_TYPE_LEN) + LMCS_APS);
          if (lmcsAPS == NULL)
          {
            lmcsAPS = apsMap->allocatePS((apsId << NUM_APS_TYPE_LEN) + LMCS_APS);
            lmcsAPS->setAPSId(apsId);
            lmcsAPS->setAPSType(LMCS_APS);
          }
          pcSlice->setLmcsAPS(lmcsAPS);
        }
        //m_pcReshaper->copySliceReshaperInfo(lmcsAPS->getReshaperAPSInfo(), m_pcReshaper->getSliceReshaperInfo());
        SliceReshapeInfo& tInfo = lmcsAPS->getReshaperAPSInfo();
        SliceReshapeInfo& sInfo = m_pcReshaper->getSliceReshaperInfo();
        tInfo.reshaperModelMaxBinIdx = sInfo.reshaperModelMaxBinIdx;
        tInfo.reshaperModelMinBinIdx = sInfo.reshaperModelMinBinIdx;
        memcpy(tInfo.reshaperModelBinCWDelta, sInfo.reshaperModelBinCWDelta, sizeof(int)*(PIC_CODE_CW_BINS));
        tInfo.maxNbitsNeededDeltaCW = sInfo.maxNbitsNeededDeltaCW;
        m_pcEncLib->getApsMap()->setChangedFlag((lmcsAPS->getAPSId() << NUM_APS_TYPE_LEN) + LMCS_APS);
      }


      if (pcSlice->getLmcsEnabledFlag())
      {
        int apsId = 0;
        pcSlice->setLmcsAPSId(apsId);
      }
    }
    else
    {
      m_pcReshaper->setCTUFlag(false);
    }

    if( encPic )
    // now compress (trial encode) the various slice segments (slices, and dependent slices)
    {
      DTRACE_UPDATE( g_trace_ctx, ( std::make_pair( "poc", pocCurr ) ) );

      pcSlice->setSliceCurStartCtuTsAddr( 0 );

      uint32_t sliceIdx = 0;
      const BrickMap& tileMap = *(pcPic->brickMap);
      for(uint32_t nextCtuTsAddr = 0; nextCtuTsAddr < numberOfCtusInFrame; )
      {
        m_pcSliceEncoder->precompressSlice( pcPic );
        m_pcSliceEncoder->compressSlice   ( pcPic, false, false );

        const uint32_t curSliceEnd = pcSlice->getSliceCurEndCtuTsAddr();
        pcSlice->setSliceIndex(sliceIdx);
        if(curSliceEnd < numberOfCtusInFrame)
        {
          uint32_t independentSliceIdx = pcSlice->getIndependentSliceIdx();
          pcPic->allocateNewSlice();
          m_pcSliceEncoder->setSliceSegmentIdx      (uiNumSliceSegments);
          // prepare for next slice
          pcSlice = pcPic->slices[uiNumSliceSegments];
          CHECK(!(pcSlice->getPPS() != 0), "Unspecified error");
          pcSlice->copySliceInfo(pcPic->slices[uiNumSliceSegments - 1]);
          sliceIdx++;
          if (pcSlice->getPPS()->getRectSliceFlag())
          {
            uint32_t startTileIdx = pcSlice->getPPS()->getTopLeftBrickIdx(sliceIdx);
            uint32_t nextCtu = 0;
            uint32_t tmpSliceIdx = 0;
            while (tmpSliceIdx != startTileIdx)
            {
              nextCtu++;
              tmpSliceIdx = tileMap.getBrickIdxBsMap(nextCtu);
            }
            pcSlice->setSliceCurStartCtuTsAddr(nextCtu);
          }
          else
          {
            pcSlice->setSliceCurStartCtuTsAddr(curSliceEnd);
          }
          pcSlice->setSliceBits(0);
          independentSliceIdx++;
          pcSlice->setIndependentSliceIdx(independentSliceIdx);
          uiNumSliceSegments++;
        }
        nextCtuTsAddr = curSliceEnd;
      }

      duData.clear();

      CodingStructure& cs = *pcPic->cs;
      pcSlice = pcPic->slices[0];

      if (pcSlice->getSPS()->getUseReshaper() && m_pcReshaper->getSliceReshaperInfo().getUseSliceReshaper())
      {
        pcSlice->setLmcsEnabledFlag(true);
        int apsId = 0;
        pcSlice->setLmcsAPSId(apsId);
        for (int s = 0; s < uiNumSliceSegments; s++)
        {
          pcPic->slices[s]->setLmcsEnabledFlag(pcSlice->getLmcsEnabledFlag());
          pcPic->slices[s]->setLmcsChromaResidualScaleFlag((pcSlice->getLmcsChromaResidualScaleFlag()));
          if (pcSlice->getLmcsEnabledFlag())
          {
            //pcPic->slices[s]->setLmcsAPS(pcSlice->getLmcsAPS());
            pcPic->slices[s]->setLmcsAPSId(pcSlice->getLmcsAPSId());
          }
        }
          CHECK((m_pcReshaper->getRecReshaped() == false), "Rec picture is not reshaped!");
          pcPic->getRecoBuf(COMPONENT_Y).rspSignal(m_pcReshaper->getInvLUT());
          m_pcReshaper->setRecReshaped(false);

          pcPic->getOrigBuf().copyFrom(pcPic->getTrueOrigBuf());
      }

      // SAO parameter estimation using non-deblocked pixels for CTU bottom and right boundary areas
      if( pcSlice->getSPS()->getSAOEnabledFlag() && m_pcCfg->getSaoCtuBoundary() )
      {
        m_pcSAO->getPreDBFStatistics( cs );
      }

      //-- Loop filter
      if ( m_pcCfg->getDeblockingFilterMetric() )
      {
  #if W0038_DB_OPT
        if ( m_pcCfg->getDeblockingFilterMetric()==2 )
        {
          applyDeblockingFilterParameterSelection(pcPic, uiNumSliceSegments, iGOPid);
        }
        else
        {
  #endif
          applyDeblockingFilterMetric(pcPic, uiNumSliceSegments);
  #if W0038_DB_OPT
        }
  #endif
      }

      m_pcLoopFilter->loopFilterPic( cs );

      CS::setRefinedMotionField(cs);
      DTRACE_UPDATE( g_trace_ctx, ( std::make_pair( "final", 1 ) ) );

      if( pcSlice->getSPS()->getSAOEnabledFlag() )
      {
        bool sliceEnabled[MAX_NUM_COMPONENT];
        m_pcSAO->initCABACEstimator( m_pcEncLib->getCABACEncoder(), m_pcEncLib->getCtxCache(), pcSlice );

        m_pcSAO->SAOProcess( cs, sliceEnabled, pcSlice->getLambdas(),
#if ENABLE_QPA
                             (m_pcCfg->getUsePerceptQPA() && !m_pcCfg->getUseRateCtrl() && pcSlice->getPPS()->getUseDQP() ? m_pcEncLib->getRdCost (PARL_PARAM0 (0))->getChromaWeight() : 0.0),
#endif
#if K0238_SAO_GREEDY_MERGE_ENCODING
                             m_pcCfg->getTestSAODisableAtPictureLevel(), m_pcCfg->getSaoEncodingRate(), m_pcCfg->getSaoEncodingRateChroma(), m_pcCfg->getSaoCtuBoundary(), m_pcCfg->getSaoGreedyMergeEnc() );
#else
                             m_pcCfg->getTestSAODisableAtPictureLevel(), m_pcCfg->getSaoEncodingRate(), m_pcCfg->getSaoEncodingRateChroma(), m_pcCfg->getSaoCtuBoundary() );
#endif
        //assign SAO slice header
        for(int s=0; s< uiNumSliceSegments; s++)
        {
          pcPic->slices[s]->setSaoEnabledFlag(CHANNEL_TYPE_LUMA, sliceEnabled[COMPONENT_Y]);
          CHECK(!(sliceEnabled[COMPONENT_Cb] == sliceEnabled[COMPONENT_Cr]), "Unspecified error");
          pcPic->slices[s]->setSaoEnabledFlag(CHANNEL_TYPE_CHROMA, sliceEnabled[COMPONENT_Cb]);
        }
      }

      if( pcSlice->getSPS()->getALFEnabledFlag() )
      {
        for (int s = 0; s < uiNumSliceSegments; s++)
        {
          pcPic->slices[s]->setTileGroupAlfEnabledFlag(COMPONENT_Y, false);
        }
        m_pcALF->initCABACEstimator(m_pcEncLib->getCABACEncoder(), m_pcEncLib->getCtxCache(), pcSlice, m_pcEncLib->getApsMap());
        m_pcALF->ALFProcess(cs, pcSlice->getLambdas()
#if ENABLE_QPA
          , (m_pcCfg->getUsePerceptQPA() && !m_pcCfg->getUseRateCtrl() && pcSlice->getPPS()->getUseDQP() ? m_pcEncLib->getRdCost(PARL_PARAM0(0))->getChromaWeight() : 0.0)
#endif
        );

        //assign ALF slice header
        for (int s = 0; s < uiNumSliceSegments; s++)
        {
          pcPic->slices[s]->setTileGroupAlfEnabledFlag(COMPONENT_Y, cs.slice->getTileGroupAlfEnabledFlag(COMPONENT_Y));
          pcPic->slices[s]->setTileGroupAlfEnabledFlag(COMPONENT_Cb, cs.slice->getTileGroupAlfEnabledFlag(COMPONENT_Cb));
          pcPic->slices[s]->setTileGroupAlfEnabledFlag(COMPONENT_Cr, cs.slice->getTileGroupAlfEnabledFlag(COMPONENT_Cr));
          if (pcPic->slices[s]->getTileGroupAlfEnabledFlag(COMPONENT_Y))
          {
            pcPic->slices[s]->setTileGroupNumAps(cs.slice->getTileGroupNumAps());
            pcPic->slices[s]->setAlfAPSs(cs.slice->getTileGroupApsIdLuma());
          }
          else
          {
            pcPic->slices[s]->setTileGroupNumAps(0);
          }
          pcPic->slices[s]->setAlfAPSs(cs.slice->getAlfAPSs());
          pcPic->slices[s]->setTileGroupApsIdChroma(cs.slice->getTileGroupApsIdChroma());
        }
      }
      if (m_pcCfg->getUseCompositeRef() && getPrepareLTRef())
      {
        updateCompositeReference(pcSlice, rcListPic, pocCurr);
      }
    }
    else // skip enc picture
    {
      pcSlice->setSliceQpBase( pcSlice->getSliceQp() );

#if ENABLE_QPA
      if (m_pcCfg->getUsePerceptQPA() && !m_pcCfg->getUseRateCtrl() && pcSlice->getPPS()->getUseDQP())
      {
        const double picLambda = pcSlice->getLambdas()[0];

        for (uint32_t ctuRsAddr = 0; ctuRsAddr < numberOfCtusInFrame; ctuRsAddr++)
        {
          pcPic->m_uEnerHpCtu[ctuRsAddr] = picLambda;  // initialize to slice lambda (just for safety)
        }
      }
#endif
      if( pcSlice->getSPS()->getSAOEnabledFlag() )
      {
        m_pcSAO->disabledRate( *pcPic->cs, pcPic->getSAO(1), m_pcCfg->getSaoEncodingRate(), m_pcCfg->getSaoEncodingRateChroma());
      }
    }

    if( m_pcCfg->getUseAMaxBT() )
    {
      for( const CodingUnit *cu : pcPic->cs->cus )
      {
        if( !pcSlice->isIRAP() )
        {
          m_uiBlkSize[pcSlice->getDepth()] += cu->Y().area();
          m_uiNumBlk [pcSlice->getDepth()]++;
        }
      }
    }

    if( encPic || decPic )
    {
      pcSlice = pcPic->slices[0];

      /////////////////////////////////////////////////////////////////////////////////////////////////// File writing

      // write various parameter sets
      bool writePS = m_bSeqFirst || (m_pcCfg->getReWriteParamSets() && (pcSlice->isIRAP()));
      if (writePS)
      {
        m_pcEncLib->setParamSetChanged(pcSlice->getSPS()->getSPSId(), pcSlice->getPPS()->getPPSId());
      }
      actualTotalBits += xWriteParameterSets(accessUnit, pcSlice, writePS);

      if (writePS)
      {
        // create prefix SEI messages at the beginning of the sequence
        CHECK(!(leadingSeiMessages.empty()), "Unspecified error");
        xCreateIRAPLeadingSEIMessages(leadingSeiMessages, pcSlice->getSPS(), pcSlice->getPPS());

        m_bSeqFirst = false;
      }
      if (m_pcCfg->getAccessUnitDelimiter())
      {
        xWriteAccessUnitDelimiter(accessUnit, pcSlice);
      }

      //send LMCS APS when LMCSModel is updated. It can be updated even current slice does not enable reshaper.
      //For example, in RA, update is on intra slice, but intra slice may not use reshaper
      if (pcSlice->getSPS()->getUseReshaper())
      {
        //only 1 LMCS data for 1 picture
        int apsId = pcSlice->getLmcsAPSId();
        ParameterSetMap<APS> *apsMap = m_pcEncLib->getApsMap();
        APS* aps = apsMap->getPS((apsId << NUM_APS_TYPE_LEN) + LMCS_APS);
        bool writeAPS = aps && apsMap->getChangedFlag((apsId << NUM_APS_TYPE_LEN) + LMCS_APS);
        if (writeAPS)
        {
          actualTotalBits += xWriteAPS(accessUnit, aps);
          apsMap->clearChangedFlag((apsId << NUM_APS_TYPE_LEN) + LMCS_APS);
          CHECK(aps != pcSlice->getLmcsAPS(), "Wrong LMCS APS pointer in compressGOP");
        }
      }

      if (pcSlice->getSPS()->getALFEnabledFlag() && pcSlice->getTileGroupAlfEnabledFlag(COMPONENT_Y))
      {
        for (int apsId = 0; apsId < MAX_NUM_APS; apsId++)   //HD: shouldn't this be looping over slice_alf_aps_id_luma[ i ]? By looping over MAX_NUM_APS, it is possible unused ALF APS is written. Please check!
        {
          ParameterSetMap<APS> *apsMap = m_pcEncLib->getApsMap();
          APS* aps = apsMap->getPS((apsId << NUM_APS_TYPE_LEN) + ALF_APS);
          bool writeAPS = aps && apsMap->getChangedFlag((apsId << NUM_APS_TYPE_LEN) + ALF_APS);
          if (!aps && pcSlice->getAlfAPSs() && pcSlice->getAlfAPSs()[apsId])
          {
            writeAPS = true;
            aps = pcSlice->getAlfAPSs()[apsId]; // use asp from slice header
            *apsMap->allocatePS(apsId) = *aps; //allocate and cpy
        }
          
          if (writeAPS )
          {
            actualTotalBits += xWriteAPS(accessUnit, aps);
            apsMap->clearChangedFlag((apsId << NUM_APS_TYPE_LEN) + ALF_APS);
            CHECK(aps != pcSlice->getAlfAPSs()[apsId], "Wrong APS pointer in compressGOP");
          }
        }
      }

      // reset presence of BP SEI indication
      m_bufferingPeriodSEIPresentInAU = false;
      // create prefix SEI associated with a picture
      xCreatePerPictureSEIMessages(iGOPid, leadingSeiMessages, nestedSeiMessages, pcSlice);

      // pcSlice is currently slice 0.
      std::size_t binCountsInNalUnits   = 0; // For implementation of cabac_zero_word stuffing (section 7.4.3.10)
      std::size_t numBytesInVclNalUnits = 0; // For implementation of cabac_zero_word stuffing (section 7.4.3.10)

      for(uint32_t sliceSegmentStartCtuTsAddr = 0, sliceSegmentIdxCount = 0; sliceSegmentStartCtuTsAddr < numberOfCtusInFrame; sliceSegmentIdxCount++, sliceSegmentStartCtuTsAddr = pcSlice->getSliceCurEndCtuTsAddr())
      {
        pcSlice = pcPic->slices[sliceSegmentIdxCount];
        if(sliceSegmentIdxCount > 0 && pcSlice->getSliceType()!= I_SLICE)
        {
          pcSlice->checkColRefIdx(sliceSegmentIdxCount, pcPic);
        }
        m_pcSliceEncoder->setSliceSegmentIdx(sliceSegmentIdxCount);

        pcSlice->setRPL0(pcPic->slices[0]->getRPL0());
        pcSlice->setRPL1(pcPic->slices[0]->getRPL1());
        pcSlice->setRPL0idx(pcPic->slices[0]->getRPL0idx());
        pcSlice->setRPL1idx(pcPic->slices[0]->getRPL1idx());

        for ( uint32_t ui = 0 ; ui < numSubstreams; ui++ )
        {
          substreamsOut[ui].clear();
        }

        /* start slice NALunit */
        OutputNALUnit nalu( pcSlice->getNalUnitType(), pcSlice->getTLayer() );
        m_HLSWriter->setBitstream( &nalu.m_Bitstream );

        pcSlice->setNoRaslOutputFlag(false);
        if (pcSlice->isIRAP())
        {
          if (pcSlice->getNalUnitType() >= NAL_UNIT_CODED_SLICE_IDR_W_RADL && pcSlice->getNalUnitType() <= NAL_UNIT_CODED_SLICE_IDR_N_LP)
          {
            pcSlice->setNoRaslOutputFlag(true);
          }
          //the inference for NoOutputPriorPicsFlag
          // KJS: This cannot happen at the encoder
          if (!m_bFirst && pcSlice->isIRAP() && pcSlice->getNoRaslOutputFlag())
          {
            if (pcSlice->getNalUnitType() == NAL_UNIT_CODED_SLICE_CRA)
            {
              pcSlice->setNoOutputPriorPicsFlag(true);
            }
          }
        }

        tmpBitsBeforeWriting = m_HLSWriter->getNumberOfWrittenBits();
        m_HLSWriter->codeSliceHeader( pcSlice );
        actualHeadBits += ( m_HLSWriter->getNumberOfWrittenBits() - tmpBitsBeforeWriting );

        pcSlice->setFinalized(true);

        pcSlice->clearSubstreamSizes(  );
        {
          uint32_t numBinsCoded = 0;
          m_pcSliceEncoder->encodeSlice(pcPic, &(substreamsOut[0]), numBinsCoded);
          binCountsInNalUnits+=numBinsCoded;
        }
        {
          // Construct the final bitstream by concatenating substreams.
          // The final bitstream is either nalu.m_Bitstream or pcBitstreamRedirect;
          // Complete the slice header info.
          m_HLSWriter->setBitstream( &nalu.m_Bitstream );
          m_HLSWriter->codeTilesWPPEntryPoint( pcSlice );

          // Append substreams...
          OutputBitstream *pcOut = pcBitstreamRedirect;
          const int numSubstreamsToCode  = pcSlice->getNumberOfSubstreamSizes()+1;
          for ( uint32_t ui = 0 ; ui < numSubstreamsToCode; ui++ )
          {
            pcOut->addSubstream(&(substreamsOut[ui]));
          }
        }

        // If current NALU is the first NALU of slice (containing slice header) and more NALUs exist (due to multiple dependent slices) then buffer it.
        // If current NALU is the last NALU of slice and a NALU was buffered, then (a) Write current NALU (b) Update an write buffered NALU at approproate location in NALU list.
        bool bNALUAlignedWrittenToList    = false; // used to ensure current NALU is not written more than once to the NALU list.
        xAttachSliceDataToNalUnit(nalu, pcBitstreamRedirect);
        accessUnit.push_back(new NALUnitEBSP(nalu));
        actualTotalBits += uint32_t(accessUnit.back()->m_nalUnitData.str().size()) * 8;
        numBytesInVclNalUnits += (std::size_t)(accessUnit.back()->m_nalUnitData.str().size());
        bNALUAlignedWrittenToList = true;

        if (!bNALUAlignedWrittenToList)
        {
          nalu.m_Bitstream.writeAlignZero();
          accessUnit.push_back(new NALUnitEBSP(nalu));
        }

        if( ( m_pcCfg->getPictureTimingSEIEnabled() || m_pcCfg->getDecodingUnitInfoSEIEnabled() ) &&
            ( pcSlice->getSPS()->getVuiParametersPresentFlag() ) &&
            ( ( pcSlice->getSPS()->getHrdParameters()->getNalHrdParametersPresentFlag() )
           || ( pcSlice->getSPS()->getHrdParameters()->getVclHrdParametersPresentFlag() ) ) &&
            ( pcSlice->getSPS()->getHrdParameters()->getSubPicCpbParamsPresentFlag() ) )
        {
            uint32_t numNalus = 0;
          uint32_t numRBSPBytes = 0;
          for (AccessUnit::const_iterator it = accessUnit.begin(); it != accessUnit.end(); it++)
          {
            numRBSPBytes += uint32_t((*it)->m_nalUnitData.str().size());
            numNalus ++;
          }
          duData.push_back(DUData());
          duData.back().accumBitsDU = ( numRBSPBytes << 3 );
          duData.back().accumNalsDU = numNalus;
        }
      } // end iteration over slices


      // cabac_zero_words processing
      cabac_zero_word_padding(pcSlice, pcPic, binCountsInNalUnits, numBytesInVclNalUnits, accessUnit.back()->m_nalUnitData, m_pcCfg->getCabacZeroWordPaddingEnabled());

      //-- For time output for each slice
      auto elapsed = std::chrono::steady_clock::now() - beforeTime;
      auto encTime = std::chrono::duration_cast<std::chrono::seconds>( elapsed ).count();

      std::string digestStr;
      if (m_pcCfg->getDecodedPictureHashSEIType()!=HASHTYPE_NONE)
      {
        SEIDecodedPictureHash *decodedPictureHashSei = new SEIDecodedPictureHash();
        PelUnitBuf recoBuf = pcPic->cs->getRecoBuf();
        m_seiEncoder.initDecodedPictureHashSEI(decodedPictureHashSei, recoBuf, digestStr, pcSlice->getSPS()->getBitDepths());
        trailingSeiMessages.push_back(decodedPictureHashSei);
      }

      m_pcCfg->setEncodedFlag(iGOPid, true);

      double PSNR_Y;
      xCalculateAddPSNRs(isField, isTff, iGOPid, pcPic, accessUnit, rcListPic, encTime, snr_conversion, printFrameMSE, &PSNR_Y
                       , isEncodeLtRef
      );

      // Only produce the Green Metadata SEI message with the last picture.
      if( m_pcCfg->getSEIGreenMetadataInfoSEIEnable() && pcSlice->getPOC() == ( m_pcCfg->getFramesToBeEncoded() - 1 )  )
      {
        SEIGreenMetadataInfo *seiGreenMetadataInfo = new SEIGreenMetadataInfo;
        m_seiEncoder.initSEIGreenMetadataInfo(seiGreenMetadataInfo, (uint32_t)(PSNR_Y * 100 + 0.5));
        trailingSeiMessages.push_back(seiGreenMetadataInfo);
      }

      xWriteTrailingSEIMessages(trailingSeiMessages, accessUnit, pcSlice->getTLayer(), pcSlice->getSPS());

      printHash(m_pcCfg->getDecodedPictureHashSEIType(), digestStr);

      if ( m_pcCfg->getUseRateCtrl() )
      {
        double avgQP     = m_pcRateCtrl->getRCPic()->calAverageQP();
        double avgLambda = m_pcRateCtrl->getRCPic()->calAverageLambda();
        if ( avgLambda < 0.0 )
        {
          avgLambda = lambda;
        }

        m_pcRateCtrl->getRCPic()->updateAfterPicture( actualHeadBits, actualTotalBits, avgQP, avgLambda, pcSlice->isIRAP());
        m_pcRateCtrl->getRCPic()->addToPictureLsit( m_pcRateCtrl->getPicList() );

        m_pcRateCtrl->getRCSeq()->updateAfterPic( actualTotalBits );
        if ( !pcSlice->isIRAP() )
        {
          m_pcRateCtrl->getRCGOP()->updateAfterPicture( actualTotalBits );
        }
        else    // for intra picture, the estimated bits are used to update the current status in the GOP
        {
          m_pcRateCtrl->getRCGOP()->updateAfterPicture( estimatedBits );
        }
  #if U0132_TARGET_BITS_SATURATION
        if (m_pcRateCtrl->getCpbSaturationEnabled())
        {
          m_pcRateCtrl->updateCpbState(actualTotalBits);
          msg( NOTICE, " [CPB %6d bits]", m_pcRateCtrl->getCpbState() );
        }
  #endif
      }

      xCreatePictureTimingSEI( m_pcCfg->getEfficientFieldIRAPEnabled() ? effFieldIRAPMap.GetIRAPGOPid() : 0, leadingSeiMessages, nestedSeiMessages, duInfoSeiMessages, pcSlice, isField, duData );
      if( m_pcCfg->getScalableNestingSEIEnabled() )
      {
        xCreateScalableNestingSEI( leadingSeiMessages, nestedSeiMessages );
      }
      xWriteLeadingSEIMessages( leadingSeiMessages, duInfoSeiMessages, accessUnit, pcSlice->getTLayer(), pcSlice->getSPS(), duData );
      xWriteDuSEIMessages( duInfoSeiMessages, accessUnit, pcSlice->getTLayer(), pcSlice->getSPS(), duData );

      m_AUWriterIf->outputAU( accessUnit );

      msg( NOTICE, "\n" );
      fflush( stdout );
    }


    DTRACE_UPDATE( g_trace_ctx, ( std::make_pair( "final", 0 ) ) );

    pcPic->reconstructed = true;
    m_bFirst = false;
    m_iNumPicCoded++;
    if (!(m_pcCfg->getUseCompositeRef() && isEncodeLtRef))
      m_totalCoded ++;
    /* logging: insert a newline at end of picture period */

    if (m_pcCfg->getEfficientFieldIRAPEnabled())
    {
      iGOPid=effFieldIRAPMap.restoreGOPid(iGOPid);
    }

    pcPic->destroyTempBuffers();
    pcPic->cs->destroyCoeffs();
    pcPic->cs->releaseIntermediateData();
  } // iGOPid-loop

  delete pcBitstreamRedirect;

  CHECK(!( (m_iNumPicCoded == iNumPicRcvd) ), "Unspecified error");

}

void EncGOP::printOutSummary(uint32_t uiNumAllPicCoded, bool isField, const bool printMSEBasedSNR, const bool printSequenceMSE, const bool printHexPsnr, const BitDepths &bitDepths)
{
#if ENABLE_QPA
  const bool    useWPSNR = m_pcEncLib->getUseWPSNR();
#endif
#if WCG_WPSNR
  const bool    useLumaWPSNR = m_pcEncLib->getLumaLevelToDeltaQPMapping().isEnabled() || (m_pcCfg->getReshaper() && m_pcCfg->getReshapeSignalType() == RESHAPE_SIGNAL_PQ);
#endif

  if( m_pcCfg->getDecodeBitstream(0).empty() && m_pcCfg->getDecodeBitstream(1).empty() && !m_pcCfg->useFastForwardToPOC() )
  {
    CHECK( !( uiNumAllPicCoded == m_gcAnalyzeAll.getNumPic() ), "Unspecified error" );
  }

  //--CFG_KDY
  const int rateMultiplier=(isField?2:1);
  m_gcAnalyzeAll.setFrmRate( m_pcCfg->getFrameRate()*rateMultiplier / (double)m_pcCfg->getTemporalSubsampleRatio());
  m_gcAnalyzeI.setFrmRate( m_pcCfg->getFrameRate()*rateMultiplier / (double)m_pcCfg->getTemporalSubsampleRatio());
  m_gcAnalyzeP.setFrmRate( m_pcCfg->getFrameRate()*rateMultiplier / (double)m_pcCfg->getTemporalSubsampleRatio());
  m_gcAnalyzeB.setFrmRate( m_pcCfg->getFrameRate()*rateMultiplier / (double)m_pcCfg->getTemporalSubsampleRatio());
#if WCG_WPSNR
  if (useLumaWPSNR)
  {
    m_gcAnalyzeWPSNR.setFrmRate(m_pcCfg->getFrameRate()*rateMultiplier / (double)m_pcCfg->getTemporalSubsampleRatio());
  }
#endif

  const ChromaFormat chFmt = m_pcCfg->getChromaFormatIdc();

  //-- all
  msg( INFO, "\n" );
  msg( DETAILS,"\nSUMMARY --------------------------------------------------------\n" );
#if ENABLE_QPA
  m_gcAnalyzeAll.printOut('a', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths, useWPSNR);
#else
  m_gcAnalyzeAll.printOut('a', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths);
#endif
  msg( DETAILS,"\n\nI Slices--------------------------------------------------------\n" );
  m_gcAnalyzeI.printOut('i', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths);

  msg( DETAILS,"\n\nP Slices--------------------------------------------------------\n" );
  m_gcAnalyzeP.printOut('p', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths);

  msg( DETAILS,"\n\nB Slices--------------------------------------------------------\n" );
  m_gcAnalyzeB.printOut('b', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths);

#if WCG_WPSNR
  if (useLumaWPSNR)
  {
    msg(DETAILS, "\nWPSNR SUMMARY --------------------------------------------------------\n");
    m_gcAnalyzeWPSNR.printOut('w', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths, useLumaWPSNR);
  }
#endif
  if (!m_pcCfg->getSummaryOutFilename().empty())
  {
    m_gcAnalyzeAll.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryOutFilename());
  }

  if (!m_pcCfg->getSummaryPicFilenameBase().empty())
  {
    m_gcAnalyzeI.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryPicFilenameBase()+"I.txt");
    m_gcAnalyzeP.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryPicFilenameBase()+"P.txt");
    m_gcAnalyzeB.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryPicFilenameBase()+"B.txt");
  }

#if WCG_WPSNR
  if (!m_pcCfg->getSummaryOutFilename().empty() && useLumaWPSNR)
  {
    m_gcAnalyzeWPSNR.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryOutFilename());
  }
#endif
  if(isField)
  {
    //-- interlaced summary
    m_gcAnalyzeAll_in.setFrmRate( m_pcCfg->getFrameRate() / (double)m_pcCfg->getTemporalSubsampleRatio());
    m_gcAnalyzeAll_in.setBits(m_gcAnalyzeAll.getBits());
    // prior to the above statement, the interlace analyser does not contain the correct total number of bits.

    msg( DETAILS,"\n\nSUMMARY INTERLACED ---------------------------------------------\n" );
#if ENABLE_QPA
    m_gcAnalyzeAll_in.printOut('a', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths, useWPSNR);
#else
    m_gcAnalyzeAll_in.printOut('a', chFmt, printMSEBasedSNR, printSequenceMSE, printHexPsnr, bitDepths);
#endif
    if (!m_pcCfg->getSummaryOutFilename().empty())
    {
      m_gcAnalyzeAll_in.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryOutFilename());
#if WCG_WPSNR
      if (useLumaWPSNR)
      {
        m_gcAnalyzeWPSNR.printSummary(chFmt, printSequenceMSE, printHexPsnr, bitDepths, m_pcCfg->getSummaryOutFilename());
      }
#endif
    }
  }

  msg( DETAILS,"\nRVM: %.3lf\n", xCalculateRVM() );
}

#if W0038_DB_OPT
uint64_t EncGOP::preLoopFilterPicAndCalcDist( Picture* pcPic )
{
  CodingStructure& cs = *pcPic->cs;
  m_pcLoopFilter->loopFilterPic( cs );

  const CPelUnitBuf picOrg = pcPic->getRecoBuf();
  const CPelUnitBuf picRec = cs.getRecoBuf();

  uint64_t uiDist = 0;
  for( uint32_t comp = 0; comp < (uint32_t)picRec.bufs.size(); comp++)
  {
    const ComponentID compID = ComponentID(comp);
    const uint32_t rshift = 2 * DISTORTION_PRECISION_ADJUSTMENT(cs.sps->getBitDepth(toChannelType(compID)));
#if ENABLE_QPA
    CHECK( rshift >= 8, "shifts greater than 7 are not supported." );
#endif
    uiDist += xFindDistortionPlane( picOrg.get(compID), picRec.get(compID), rshift );
  }
  return uiDist;
}
#endif

// ====================================================================================================================
// Protected member functions
// ====================================================================================================================
void EncGOP::xInitGOP( int iPOCLast, int iNumPicRcvd, bool isField
  , bool isEncodeLtRef
)
{
  CHECK(!( iNumPicRcvd > 0 ), "Unspecified error");
  //  Exception for the first frames
  if ((isField && (iPOCLast == 0 || iPOCLast == 1)) || (!isField && (iPOCLast == 0)) || isEncodeLtRef)
  {
    m_iGopSize    = 1;
  }
  else
  {
    m_iGopSize    = m_pcCfg->getGOPSize();
  }
  CHECK(!(m_iGopSize > 0), "Unspecified error");

  return;
}


void EncGOP::xGetBuffer( PicList&                  rcListPic,
                         std::list<PelUnitBuf*>&   rcListPicYuvRecOut,
                         int                       iNumPicRcvd,
                         int                       iTimeOffset,
                         Picture*&                 rpcPic,
                         int                       pocCurr,
                         bool                      isField )
{
  int i;
  //  Rec. output
  std::list<PelUnitBuf*>::iterator     iterPicYuvRec = rcListPicYuvRecOut.end();

  if (isField && pocCurr > 1 && m_iGopSize!=1)
  {
    iTimeOffset--;
  }

  int multipleFactor = m_pcCfg->getUseCompositeRef() ? 2 : 1;
  for (i = 0; i < (iNumPicRcvd * multipleFactor - iTimeOffset + 1); i += multipleFactor)
  {
    iterPicYuvRec--;
  }

  //  Current pic.
  PicList::iterator        iterPic       = rcListPic.begin();
  while (iterPic != rcListPic.end())
  {
    rpcPic = *(iterPic);
    if (rpcPic->getPOC() == pocCurr)
    {
      break;
    }
    iterPic++;
  }

  CHECK(!(rpcPic != NULL), "Unspecified error");
  CHECK(!(rpcPic->getPOC() == pocCurr), "Unspecified error");

  (**iterPicYuvRec) = rpcPic->getRecoBuf();
  return;
}

#if ENABLE_QPA

#ifndef BETA
  #define BETA 0.5 // value between 0.0 and 1; use 0.0 to obtain traditional PSNR
#endif

static inline double calcWeightedSquaredError(const CPelBuf& org,        const CPelBuf& rec,
                                              double &sumAct,            const uint32_t bitDepth,
                                              const uint32_t imageWidth, const uint32_t imageHeight,
                                              const uint32_t offsetX,    const uint32_t offsetY,
                                              int blockWidth,            int blockHeight)
{
  const int    O = org.stride;
  const int    R = rec.stride;
  const Pel   *o = org.bufAt(offsetX, offsetY);
  const Pel   *r = rec.bufAt(offsetX, offsetY);
  const int yAct = offsetY > 0 ? 0 : 1;
  const int xAct = offsetX > 0 ? 0 : 1;

  if (offsetY + (uint32_t)blockHeight > imageHeight) blockHeight = imageHeight - offsetY;
  if (offsetX + (uint32_t)blockWidth  > imageWidth ) blockWidth  = imageWidth  - offsetX;

  const int hAct = offsetY + (uint32_t)blockHeight < imageHeight ? blockHeight : blockHeight - 1;
  const int wAct = offsetX + (uint32_t)blockWidth  < imageWidth  ? blockWidth  : blockWidth  - 1;
  uint64_t ssErr = 0; // sum of squared diffs
  uint64_t saAct = 0; // sum of abs. activity
  double msAct;
  int x, y;

  // calculate image differences and activity
  for (y = 0; y < blockHeight; y++)  // error
  {
    for (x = 0; x < blockWidth; x++)
    {
      const     int64_t iDiff = (int64_t)o[y*O + x] - (int64_t)r[y*R + x];
      ssErr += uint64_t(iDiff * iDiff);
    }
  }
  if (wAct <= xAct || hAct <= yAct) return (double)ssErr;

  for (y = yAct; y < hAct; y++)   // activity
  {
    for (x = xAct; x < wAct; x++)
    {
      const int f = 12 * (int)o[y*O + x] - 2 * ((int)o[y*O + x-1] + (int)o[y*O + x+1] + (int)o[(y-1)*O + x] + (int)o[(y+1)*O + x])
                       - (int)o[(y-1)*O + x-1] - (int)o[(y-1)*O + x+1] - (int)o[(y+1)*O + x-1] - (int)o[(y+1)*O + x+1];
      saAct += abs(f);
    }
  }

  // calculate weight (mean squared activity)
  msAct = (double)saAct / (double(wAct - xAct) * double(hAct - yAct));

  // lower limit, accounts for high-pass gain
  if (msAct < double(1 << (bitDepth - 4))) msAct = double(1 << (bitDepth - 4));

  msAct *= msAct; // because ssErr is squared

  sumAct += msAct; // includes high-pass gain

  // calculate activity weighted error square
  return (double)ssErr * pow(msAct, -1.0 * BETA);
}
#endif // ENABLE_QPA

uint64_t EncGOP::xFindDistortionPlane(const CPelBuf& pic0, const CPelBuf& pic1, const uint32_t rshift
#if ENABLE_QPA
                                    , const uint32_t chromaShift /*= 0*/
#endif
                                      )
{
  uint64_t uiTotalDiff;
  const  Pel*  pSrc0 = pic0.bufAt(0, 0);
  const  Pel*  pSrc1 = pic1.bufAt(0, 0);

  CHECK(pic0.width  != pic1.width , "Unspecified error");
  CHECK(pic0.height != pic1.height, "Unspecified error");

  if( rshift > 0 )
  {
#if ENABLE_QPA
    const   uint32_t  BD = rshift;      // image bit-depth
    if (BD >= 8)
    {
      const uint32_t   W = pic0.width;  // image width
      const uint32_t   H = pic0.height; // image height
      const double     R = double(W * H) / (1920.0 * 1080.0);
      const uint32_t   B = Clip3<uint32_t>(0, 128 >> chromaShift, 4 * uint32_t(16.0 * sqrt(R) + 0.5)); // WPSNR block size in integer multiple of 4 (for SIMD, = 64 at full-HD)

      uint32_t x, y;

      if (B < 4) // image is too small to use WPSNR, resort to traditional PSNR
      {
        uiTotalDiff = 0;
        for (y = 0; y < H; y++)
        {
          for (x = 0; x < W; x++)
          {
            const           int64_t iDiff = (int64_t)pSrc0[x] - (int64_t)pSrc1[x];
            uiTotalDiff += uint64_t(iDiff * iDiff);
          }
          pSrc0 += pic0.stride;
          pSrc1 += pic1.stride;
        }
        return uiTotalDiff;
      }

      double wmse = 0.0, sumAct = 0.0; // compute activity normalized SNR value

      for (y = 0; y < H; y += B)
      {
        for (x = 0; x < W; x += B)
        {
          wmse += calcWeightedSquaredError(pic1,   pic0,
                                           sumAct, BD,
                                           W,      H,
                                           x,      y,
                                           B,      B);
        }
      }

      // integer weighted distortion
      sumAct = 16.0 * sqrt ((3840.0 * 2160.0) / double((W << chromaShift) * (H << chromaShift))) * double(1 << BD);

      return (wmse <= 0.0) ? 0 : uint64_t(wmse * pow(sumAct, BETA) + 0.5);
    }
#endif // ENABLE_QPA
    uiTotalDiff = 0;
    for (int y = 0; y < pic0.height; y++)
    {
      for (int x = 0; x < pic0.width; x++)
      {
        Intermediate_Int iTemp = pSrc0[x] - pSrc1[x];
        uiTotalDiff += uint64_t((iTemp * iTemp) >> rshift);
      }
      pSrc0 += pic0.stride;
      pSrc1 += pic1.stride;
    }
  }
  else
  {
    uiTotalDiff = 0;
    for (int y = 0; y < pic0.height; y++)
    {
      for (int x = 0; x < pic0.width; x++)
      {
        Intermediate_Int iTemp = pSrc0[x] - pSrc1[x];