/work/vvenc/source/Lib/CommonLib/Quant.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     Quant.cpp
    \brief    transform and quantization class
*/

#include "Quant.h"
#include "UnitTools.h"
#include "ContextModelling.h"
#include "CodingStructure.h"
#include "dtrace_buffer.h"

#include <stdlib.h>
#include <memory.h>

//! \ingroup CommonLib
//! \{

namespace vvenc {

// ====================================================================================================================
// Constants
// ====================================================================================================================


// ====================================================================================================================
// QpParam constructor
// ====================================================================================================================
 
QpParam::QpParam(const TransformUnit& tu, const ComponentID &compID, const bool allowACTQpoffset)
{
  const ChannelType chType = toChannelType( compID );
  const SPS        &sps    = *tu.cu->cs->sps;
  const int     qpBdOffset = sps.qpBDOffset[chType];
  const bool useJQP        = isChroma( compID ) && abs( TU::getICTMode( tu ) ) == 2;
  const ComponentID jCbCr  = useJQP ? COMP_JOINT_CbCr : compID;
  
        int chromaQpOffset = 0;

  if( isChroma( compID ) )
  {
    const PPS &pps  = *tu.cu->slice->pps;
    chromaQpOffset  = pps.chromaQpOffset              [jCbCr];
    chromaQpOffset += tu.cu->slice->sliceChromaQpDelta[jCbCr];
    chromaQpOffset += pps.getChromaQpOffsetListEntry( tu.cu->chromaQpAdj ).u.offset[int( jCbCr ) - 1];
  }
  
  int baseQp;
  int qpy        = tu.cu->qp;
  //bool skip      = tu.mtsIdx[compID] == MTS_SKIP;

  if( isLuma( compID ) )
  {
    baseQp = tu.cu->qp + qpBdOffset;
  }
  else
  {
    int qpi = Clip3( -qpBdOffset, MAX_QP, qpy );
    baseQp  = sps.chromaQpMappingTable.getMappedChromaQpValue( jCbCr, qpi );
    baseQp  = Clip3( -qpBdOffset, MAX_QP, baseQp + chromaQpOffset ) + qpBdOffset;
  }

  if( allowACTQpoffset && tu.cu->colorTransform )
  {
    baseQp += DELTA_QP_ACT[jCbCr];
  }

  baseQp = Clip3( 0, MAX_QP + qpBdOffset, baseQp );

  //if( !skip )
  {
    Qps [0] = baseQp;
    pers[0] = baseQp / 6;
    rems[0] = baseQp % 6;
  }
  //else
  {
    int internalMinusInputBitDepth = sps.internalMinusInputBitDepth[chType];
    int baseQpTS           = std::max( baseQp, 4 + 6 * internalMinusInputBitDepth );

    Qps [1] = baseQpTS;
    pers[1] = baseQpTS / 6;
    rems[1] = baseQpTS % 6;
  }
}


// ====================================================================================================================
// Quant class member functions
// ====================================================================================================================
static void QuantCore(const TransformUnit tu, const ComponentID compID, const CCoeffBuf& piCoef,CoeffSigBuf piQCoef,TCoeff &uiAbsSum, int &lastScanPos,TCoeff *deltaU,const int defaultQuantisationCoefficient,const int iQBits,const int64_t iAdd,const TCoeff entropyCodingMinimum,const TCoeff entropyCodingMaximum,const bool signHiding, const TCoeff m_thrVal)
{
  CoeffCodingContext cctx( tu, compID, signHiding );

  const CompArea &rect      = tu.blocks[compID];
  const uint32_t uiWidth    = rect.width;
  const uint32_t uiHeight   = rect.height;

  /* for 422 chroma blocks, the effective scaling applied during transformation is not a power of 2, hence it cannot be
  * implemented as a bit-shift (the quantised result will be sqrt(2) * larger than required). Alternatively, adjust the
  * uiLog2TrSize applied in iTransformShift, such that the result is 1/sqrt(2) the required result (i.e. smaller)
  * Then a QP+3 (sqrt(2)) or QP-3 (1/sqrt(2)) method could be used to get the required result
  */

  const uint32_t log2CGSize         = cctx.log2CGSize();

  uiAbsSum = 0;

  const int iCGSize   = 1 << log2CGSize;

  const uint32_t lfnstIdx = tu.cu->lfnstIdx;
  const int iCGNum   = lfnstIdx > 0 ? 1 : std::min<int>(JVET_C0024_ZERO_OUT_TH, uiWidth) * std::min<int>(JVET_C0024_ZERO_OUT_TH, uiHeight) >> cctx.log2CGSize();
  int       iScanPos = ( iCGNum << log2CGSize ) - 1;

  if( lfnstIdx > 0 && ( ( uiWidth == 4 && uiHeight == 4 ) || ( uiWidth == 8 && uiHeight == 8 ) ) )
  {
    iScanPos = 7;
  }

  // Find first non-zero coeff
  for( ; iScanPos > 0; iScanPos-- )
  {
    uint32_t uiBlkPos = cctx.blockPos( iScanPos );
    if( piCoef.buf[uiBlkPos] )
      break;
  }

  //////////////////////////////////////////////////////////////////////////
  //  Loop over sub-sets (coefficient groups)
  //////////////////////////////////////////////////////////////////////////
  
  TCoeff thres = 0, useThres = 0;
  
  if( iQBits )
    thres = TCoeff( ( int64_t( m_thrVal ) << ( iQBits - 1 ) ) );
  else
    thres = TCoeff( ( int64_t( m_thrVal >> 1 ) << iQBits ) );

  useThres = thres / ( defaultQuantisationCoefficient << 2 );

  const bool is4x4sbb = log2CGSize == 4 && cctx.log2CGWidth() == 2;

  int subSetId = iScanPos >> log2CGSize;
  for( ; subSetId >= 1; subSetId-- )
  {
    if( is4x4sbb && iScanPos >= 16 )
    {
      int  iScanPosinCG = iScanPos & ( iCGSize - 1 );
      bool allSmaller   = true;

      for( int xScanPosinCG = iScanPosinCG, xScanPos = iScanPos; allSmaller && xScanPosinCG >= 0; xScanPosinCG--, xScanPos-- )
      {
        const uint32_t uiBlkPos = cctx.blockPos( xScanPos );
        allSmaller &= abs( piCoef.buf[uiBlkPos] ) <= useThres;
      }

      if( allSmaller )
      {
        iScanPos    -= iScanPosinCG + 1;
        continue;
      }
      else
      {
        break;
      }
    }
  }

  const int qBits8 = iQBits - 8;
  piQCoef.memset( 0 );

  for( int currPos = 0; currPos <= iScanPos; currPos++ )
  {
    const int uiBlockPos  = cctx.blockPos( currPos );
    const TCoeff iLevel   = piCoef.buf[uiBlockPos];
    const TCoeff iSign    = (iLevel < 0 ? -1: 1);

    const int64_t  tmpLevel = (int64_t)abs(iLevel) * defaultQuantisationCoefficient;
    const TCoeff quantisedMagnitude = TCoeff((tmpLevel + iAdd ) >> iQBits);
    deltaU[uiBlockPos] = (TCoeff)((tmpLevel - ((int64_t)quantisedMagnitude<<iQBits) )>> qBits8);

    uiAbsSum += quantisedMagnitude;
    const TCoeff quantisedCoefficient = quantisedMagnitude * iSign;

    piQCoef.buf[uiBlockPos] = Clip3<TCoeff>( entropyCodingMinimum, entropyCodingMaximum, quantisedCoefficient );
  } // for n

  lastScanPos = iScanPos;
}

static void DeQuantCore(const int maxX,const int maxY,const int scale,const TCoeffSig* const piQCoef,const size_t piQCfStride,TCoeff   *const piCoef,const int rightShift,const int inputMaximum,const TCoeff transformMaximum)
{
  const int inputMinimum = -(inputMaximum+1);
  const TCoeff transformMinimum = -(transformMaximum+1);
  if (rightShift>0)
  {
    const Intermediate_Int iAdd = (Intermediate_Int) 1 << (rightShift - 1);
    for( int y = 0, n = 0; y <= maxY; y++)
    {
      for( int x = 0; x <= maxX; x++, n++ )
      {
        const TCoeff           clipQCoef = TCoeff(Clip3<Intermediate_Int>(inputMinimum, inputMaximum, piQCoef[x + y * piQCfStride]));
        Intermediate_Int iCoeffQ   = (Intermediate_Int(clipQCoef) * scale + iAdd) >> rightShift;
        piCoef[n] = TCoeff(Clip3<Intermediate_Int>(transformMinimum,transformMaximum,iCoeffQ));
      }
    }
  }
  else  // rightshift <0
  {
    int leftShift = -rightShift;
    for( int y = 0, n = 0; y <= maxY; y++)
    {
      for( int x = 0; x <= maxX; x++, n++ )
      {
        const TCoeff           clipQCoef = TCoeff(Clip3<Intermediate_Int>(inputMinimum, inputMaximum, piQCoef[x + y * piQCfStride]));
        const Intermediate_Int iCoeffQ   = (Intermediate_Int(clipQCoef) * scale) * (1 << leftShift);
        piCoef[n] = TCoeff(Clip3<Intermediate_Int>(transformMinimum,transformMaximum,iCoeffQ));
      }
    }
  }
}

static bool needRdoqCore( const TCoeff* pCoeff, size_t numCoeff, int quantCoeff, int64_t offset, int shift )
{
  for( int uiBlockPos = 0; uiBlockPos < numCoeff; uiBlockPos++ )
  {
    const TCoeff   iLevel = pCoeff[uiBlockPos];
    const int64_t  tmpLevel = ( int64_t ) std::abs( iLevel ) * quantCoeff;
    const TCoeff quantisedMagnitude = TCoeff( ( tmpLevel + offset ) >> shift );

    if( quantisedMagnitude != 0 )
    {
      return true;
    }
  } // for n
  return false;
}


void Quant::fwdResDPCMCore( TCoeffSig* coef, const ptrdiff_t stride, const int width, const int height, const bool vertical )
{
  if( !vertical )
  {
    for( int y = 0; y < height; y++ )
    {
      for( int x = width - 1; x > 0; x-- )
      {
        coef[x] -= coef[x - 1];
      }
      coef += stride;
    }
  }
  else
  {
    coef += stride * (height - 1);
    for( int y = 0; y < height - 1; y++ )
    {
      for ( int x = 0; x < width; x++ )
      {
        coef[x] -= coef[x - stride];
      }
      coef -= stride;
    }
  }
}

void Quant::invResDPCMVerCore( const TCoeffSig* coef, const ptrdiff_t coefStride, TCoeffSig* dst, const ptrdiff_t dstStride, const int width, const int height )
{
  const TCoeff inputMinimum = std::numeric_limits<TCoeffSig>::min();
  const TCoeff inputMaximum = std::numeric_limits<TCoeffSig>::max();

  for( int x = 0; x < width; x++ )
  {
    dst[x] = coef[x];
  }
  for( int y = 0; y < height - 1; y++ )
  {
    for( int x = 0; x < width; x++ )
    {
      dst[dstStride + x] = Clip3(inputMinimum, inputMaximum, TCoeff( dst[x] ) + TCoeff( coef[coefStride + x] ));
    }
    coef += coefStride;
    dst += dstStride;
  }
}

Quant::Quant( const Quant* other, bool useScalingLists, bool enableOpt ) : m_RDOQ( 0 ), m_useRDOQTS( false ), m_dLambda( 0.0 )
{
  xInitScalingList( other, useScalingLists );
  xDeQuant       = DeQuantCore;
  xQuant         = QuantCore;
  xNeedRdoq      = needRdoqCore;
  xFwdResDPCM    = fwdResDPCMCore;
  xInvResDPCMVer = invResDPCMVerCore;
  if( enableOpt )
  {
#if defined( TARGET_SIMD_X86 ) && ENABLE_SIMD_OPT_QUANT
    initQuantX86();
#endif
#if defined( TARGET_SIMD_ARM ) && ENABLE_SIMD_OPT_QUANT
    initQuantARM();
#endif
  }
}

Quant::~Quant()
{
  xDestroyScalingList();
}

void Quant::xInvResDPCM( const TransformUnit& tu, const ComponentID compID, CoeffSigBuf& dstBuf ) const
{
  const CompArea&    rect   = tu.blocks[compID];
  const int          wdt    = rect.width;
  const int          hgt    = rect.height;
  const CCoeffSigBuf coeffs = tu.getCoeffs(compID);

  const int      maxLog2TrDynamicRange = tu.cs->sps->getMaxLog2TrDynamicRange();
  const TCoeff   inputMinimum          = -(1 << maxLog2TrDynamicRange);
  const TCoeff   inputMaximum          =  (1 << maxLog2TrDynamicRange) - 1;

  const TCoeffSig* coef = &coeffs.buf[0];
        TCoeffSig* dst  = &dstBuf.buf[0];

  if ( tu.cu->bdpcmM[toChannelType(compID)] == 1)
  {
    // a running sum along each row, clipped at every step: sequential
    for( int y = 0; y < hgt; y++ )
    {
      dst[0] = coef[0];
      for( int x = 1; x < wdt; x++ )
      {
        dst[x] = Clip3(inputMinimum, inputMaximum, TCoeff( dst[x - 1] ) + TCoeff( coef[x] ));
      }
      coef += coeffs.stride;
      dst += dstBuf.stride;
    }
  }
  else
  {
    CHECKD( maxLog2TrDynamicRange != 15, "the vertical inverse clips to the 16-bit range" );
    xInvResDPCMVer( coef, coeffs.stride, dst, dstBuf.stride, wdt, hgt );
  }
}

// To minimize the distortion only. No rate is considered.
void Quant::xSignBitHidingHDQ( TCoeffSig* pQCoef, const TCoeff* pCoef, TCoeff* deltaU, const CoeffCodingContext& cctx, int& lastScanPos, const int maxLog2TrDynamicRange )
{
  const uint32_t groupSize = 1 << cctx.log2CGSize();

  const TCoeff entropyCodingMinimum = -(1 << maxLog2TrDynamicRange);
  const TCoeff entropyCodingMaximum =  (1 << maxLog2TrDynamicRange) - 1;

  int lastCG = -1;
  int absSum = 0 ;
  int n ;

  for( int subSet = lastScanPos >> cctx.log2CGSize(); subSet >= 0; subSet-- )
  {
    int  subPos = subSet << cctx.log2CGSize();
    int  firstNZPosInCG=groupSize , lastNZPosInCG=-1 ;
    absSum = 0 ;

    for(n = groupSize-1; n >= 0; --n )
    {
      if( pQCoef[ cctx.blockPos( n + subPos ) ] )
      {
        lastNZPosInCG = n;
        break;
      }
    }

    for(n = 0; n <groupSize; n++ )
    {
      if( pQCoef[ cctx.blockPos( n + subPos ) ] )
      {
        firstNZPosInCG = n;
        break;
      }
    }

    for(n = firstNZPosInCG; n <=lastNZPosInCG; n++ )
    {
      absSum += int(pQCoef[ cctx.blockPos( n + subPos ) ]);
    }

    if(lastNZPosInCG>=0 && lastCG==-1)
    {
      lastCG = 1 ;
    }

    if( lastNZPosInCG-firstNZPosInCG>=SBH_THRESHOLD )
    {
      uint32_t signbit = (pQCoef[cctx.blockPos(subPos+firstNZPosInCG)]>0?0:1) ;
      if( signbit!=(absSum&0x1) )  //compare signbit with sum_parity
      {
        TCoeff curCost    = std::numeric_limits<TCoeff>::max();
        TCoeff minCostInc = std::numeric_limits<TCoeff>::max();
        int minPos =-1, finalChange=0, curChange=0, minScanPos = -1;

        for( n = (lastCG==1?lastNZPosInCG:groupSize-1) ; n >= 0; --n )
        {
          uint32_t blkPos   = cctx.blockPos( n+subPos );
          if(pQCoef[ blkPos ] != 0 )
          {
            if(deltaU[blkPos]>0)
            {
              curCost = - deltaU[blkPos];
              curChange=1 ;
            }
            else
            {
              //curChange =-1;
              if(n==firstNZPosInCG && abs(pQCoef[blkPos])==1)
              {
                curCost = std::numeric_limits<TCoeff>::max();
              }
              else
              {
                curCost = deltaU[blkPos];
                curChange =-1;
              }
            }
          }
          else
          {
            if(n<firstNZPosInCG)
            {
              uint32_t thisSignBit = (pCoef[blkPos]>=0?0:1);
              if(thisSignBit != signbit )
              {
                curCost = std::numeric_limits<TCoeff>::max();
              }
              else
              {
                curCost = - (deltaU[blkPos])  ;
                curChange = 1 ;
              }
            }
            else
            {
              curCost = - (deltaU[blkPos])  ;
              curChange = 1 ;
            }
          }

          if( curCost<minCostInc)
          {
            minCostInc = curCost ;
            finalChange = curChange ;
            minPos = blkPos;
            minScanPos = n + subPos;
          }
        } //CG loop

        if(pQCoef[minPos] == entropyCodingMaximum || pQCoef[minPos] == entropyCodingMinimum)
        {
          finalChange = -1;
        }

        if(pCoef[minPos]>=0)
        {
          pQCoef[minPos] += finalChange ;
        }
        else
        {
          pQCoef[minPos] -= finalChange ;
        }

        // if changing lastScanPos element to 0, move the pointer to the new lastScanPos element
        if( minScanPos == lastScanPos && pQCoef[minPos] == 0 )
        {
          for( ; lastScanPos >= 0 && pQCoef[cctx.blockPos( lastScanPos )] == 0; lastScanPos-- );
        }
        else if( minScanPos > lastScanPos && pQCoef[minPos] != 0 )
        {
          lastScanPos = minPos;
        }
      } // Hide
    }
    if(lastCG==1)
    {
      lastCG=0 ;
    }
  } // TU loop

  return;
}

void Quant::dequant(const TransformUnit& tu,
                          CoeffBuf&      dstCoeff,
                    const ComponentID    compID,
                    const QpParam&       cQP)
{
  const SPS       *sps                  = tu.cs->sps;
  const CompArea  &area                 = tu.blocks[compID];
  const uint32_t  uiWidth               = area.width;
  const uint32_t  uiHeight              = area.height;
  TCoeff *const   piCoef                = dstCoeff.buf;
  const uint32_t  numSamplesInBlock     = uiWidth * uiHeight;
  const int       maxLog2TrDynamicRange = sps->getMaxLog2TrDynamicRange();
  const TCoeff    transformMinimum      = -(1 << maxLog2TrDynamicRange);
  const TCoeff    transformMaximum      =  (1 << maxLog2TrDynamicRange) - 1;
  const bool      isTransformSkip       = tu.mtsIdx[compID] == MTS_SKIP;
  const bool      isLfnstApplied        = tu.cu->lfnstIdx > 0 && (CU::isSepTree(*tu.cu) ? true : isLuma(compID));
  const bool      enableScalingLists    = getUseScalingList(uiWidth, uiHeight, isTransformSkip, isLfnstApplied);
  const int       scalingListType       = getScalingListType(tu.cu->predMode, compID);
  const int       channelBitDepth       = sps->bitDepths[toChannelType(compID)];

  const TCoeffSig *coef     = tu.getCoeffs( compID ).buf;
  const ptrdiff_t  piStride = tu.getCoeffs( compID ).stride;

  if( tu.cu->bdpcmM[toChannelType( compID )] )
  {
    CoeffSigBuf coefBuf( m_tmpBdpcm, uiWidth, uiHeight );
    xInvResDPCM( tu, compID, coefBuf );
    coef      = m_tmpBdpcm;
  }

  const TCoeffSig  *const piQCoef = coef;
  CHECK(scalingListType >= SCALING_LIST_NUM, "Invalid scaling list");

  // Represents scaling through forward transform
  const int  originalTransformShift = getTransformShift(channelBitDepth, area.size(), maxLog2TrDynamicRange);
  const bool needSqrtAdjustment     = TU::needsSqrt2Scale( tu, compID );
  const int  iTransformShift        = originalTransformShift + (needSqrtAdjustment?-1:0);

  const int QP_per = cQP.per(isTransformSkip);
  const int QP_rem = cQP.rem(isTransformSkip);

  const int  rightShift = (IQUANT_SHIFT - ((isTransformSkip ? 0 : iTransformShift) + QP_per)) + (enableScalingLists ? LOG2_SCALING_LIST_NEUTRAL_VALUE : 0);

  if(enableScalingLists)
  {
    //from the dequantization equation:
    //iCoeffQ                         = ((Intermediate_Int(clipQCoef) * piDequantCoef[deQuantIdx]) + iAdd ) >> rightShift
    //(sizeof(Intermediate_Int) * 8)  =              inputBitDepth    +    dequantCoefBits                   - rightShift
    const uint32_t             dequantCoefBits     = 1 + IQUANT_SHIFT + SCALING_LIST_BITS;
    const uint32_t             targetInputBitDepth = std::min<uint32_t>((maxLog2TrDynamicRange + 1), (((sizeof(Intermediate_Int) * 8) + rightShift) - dequantCoefBits));

    const Intermediate_Int inputMinimum        = -(1 << (targetInputBitDepth - 1));
    const Intermediate_Int inputMaximum        =  (1 << (targetInputBitDepth - 1)) - 1;

    const uint32_t uiLog2TrWidth  = Log2(uiWidth);
    const uint32_t uiLog2TrHeight = Log2(uiHeight);
    int* piDequantCoef            = getDequantCoeff(scalingListType, QP_rem, uiLog2TrWidth, uiLog2TrHeight);

    if(rightShift > 0)
    {
      const Intermediate_Int iAdd = (Intermediate_Int) 1 << (rightShift - 1);
      for( int n = 0; n < numSamplesInBlock; n++ )
      {
        const TCoeff           clipQCoef = TCoeff(Clip3<Intermediate_Int>(inputMinimum, inputMaximum, piQCoef[n]));
        const Intermediate_Int iCoeffQ   = ((Intermediate_Int(clipQCoef) * piDequantCoef[n]) + iAdd ) >> rightShift;
        piCoef[n] = TCoeff(Clip3<Intermediate_Int>(transformMinimum,transformMaximum,iCoeffQ));
      }
    }
    else
    {
      const int leftShift = -rightShift;
      for( int n = 0; n < numSamplesInBlock; n++ )
      {
        const TCoeff           clipQCoef = TCoeff(Clip3<Intermediate_Int>(inputMinimum, inputMaximum, piQCoef[n]));
        const Intermediate_Int iCoeffQ   = (Intermediate_Int(clipQCoef) * piDequantCoef[n]) << leftShift;
        piCoef[n] = TCoeff(Clip3<Intermediate_Int>(transformMinimum,transformMaximum,iCoeffQ));
      }
    }
  }
  else
  {
    const int scale     = g_invQuantScales[needSqrtAdjustment?1:0][QP_rem];
    const int scaleBits = ( IQUANT_SHIFT + 1 );
    //from the dequantisation equation:
    //iCoeffQ                         = Intermediate_Int((int64_t(clipQCoef) * scale + iAdd) >> rightShift);
    //(sizeof(Intermediate_Int) * 8)  =                    inputBitDepth   + scaleBits      - rightShift
    const uint32_t             targetInputBitDepth = std::min<uint32_t>((maxLog2TrDynamicRange + 1), (((sizeof(Intermediate_Int) * 8) + rightShift) - scaleBits));
    const Intermediate_Int inputMaximum        =  (1 << (targetInputBitDepth - 1)) - 1;
    xDeQuant(uiWidth-1,uiHeight-1,scale,piQCoef,piStride,piCoef,rightShift,inputMaximum,transformMaximum);
  }
}

void Quant::init( int rdoq, bool bUseRDOQTS, int thrVal )
{

  // TODO: pass to init() a single variable containing (quantization) flags,
  //       instead of variables that don't have to do with this class

  m_RDOQ             = rdoq;
  m_useRDOQTS        = bUseRDOQTS;
  m_thrVal           = thrVal;
}

/** set flat matrix value to quantized coefficient
 */
void Quant::setFlatScalingList(const int maxLog2TrDynamicRange[MAX_NUM_CH], const BitDepths &bitDepths )
{
  if( !m_scalingListEnabled ) return;

  const int minimumQp = 0;
  const int maximumQp = SCALING_LIST_REM_NUM;

  for(uint32_t sizeX = 0; sizeX < SCALING_LIST_SIZE_NUM; sizeX++)
  {
    for(uint32_t sizeY = 0; sizeY < SCALING_LIST_SIZE_NUM; sizeY++)
    {
      for(uint32_t list = 0; list < SCALING_LIST_NUM; list++)
      {
        for(int qp = minimumQp; qp < maximumQp; qp++)
        {
          xSetFlatScalingList( list, sizeX, sizeY, qp );
        }
      }
    }
  }
}

/** set flat matrix value to quantized coefficient
 * \param list List ID
 * \param size size index
 * \param qp Quantization parameter
 * \param format chroma format
 */
void Quant::xSetFlatScalingList(uint32_t list, uint32_t sizeX, uint32_t sizeY, int qp )
{
  uint32_t i,num = g_scalingListSizeX[sizeX]*g_scalingListSizeX[sizeY];
  int *quantcoeff;
  int *dequantcoeff;

  const bool blockIsNotPowerOf4 = ((Log2(g_scalingListSizeX[sizeX] * g_scalingListSizeX[sizeY])) & 1) == 1;
  int quantScales    = g_quantScales   [blockIsNotPowerOf4?1:0][qp];
  int invQuantScales = g_invQuantScales[blockIsNotPowerOf4?1:0][qp] << 4;

  quantcoeff   = getQuantCoeff(list, qp, sizeX, sizeY);
  dequantcoeff = getDequantCoeff(list, qp, sizeX, sizeY);

  for(i=0;i<num;i++)
  {
    *quantcoeff++ = quantScales;
    *dequantcoeff++ = invQuantScales;
  }
}


/** initialization process of scaling list array
 */
void Quant::xInitScalingList( const Quant* other, bool useScalingLists )
{
  m_isScalingListOwner = other == nullptr;
  m_scalingListEnabled = useScalingLists;

  for(uint32_t sizeIdX = 0; sizeIdX < SCALING_LIST_SIZE_NUM; sizeIdX++)
  {
    for(uint32_t sizeIdY = 0; sizeIdY < SCALING_LIST_SIZE_NUM; sizeIdY++)
    {
      for(uint32_t qp = 0; qp < SCALING_LIST_REM_NUM; qp++)
      {
        for(uint32_t listId = 0; listId < SCALING_LIST_NUM; listId++)
        {
          if( m_isScalingListOwner )
          {
            const size_t scalingListSize = g_scalingListSizeX[sizeIdX] * g_scalingListSizeX[sizeIdY];

            m_quantCoef   [sizeIdX][sizeIdY][listId][qp] = useScalingLists ? new int[scalingListSize] : nullptr;
            m_dequantCoef [sizeIdX][sizeIdY][listId][qp] = useScalingLists ? new int[scalingListSize] : nullptr;
          }
          else
          {
            m_quantCoef   [sizeIdX][sizeIdY][listId][qp] = other->m_quantCoef   [sizeIdX][sizeIdY][listId][qp];
            m_dequantCoef [sizeIdX][sizeIdY][listId][qp] = other->m_dequantCoef [sizeIdX][sizeIdY][listId][qp];
          }
        } // listID loop
      }
    }
  }
}

/** destroy quantization matrix array
 */
void Quant::xDestroyScalingList()
{
  if( !m_isScalingListOwner ) return;

  for(uint32_t sizeIdX = 0; sizeIdX < SCALING_LIST_SIZE_NUM; sizeIdX++)
  {
    for(uint32_t sizeIdY = 0; sizeIdY < SCALING_LIST_SIZE_NUM; sizeIdY++)
    {
      for(uint32_t listId = 0; listId < SCALING_LIST_NUM; listId++)
      {
        for(uint32_t qp = 0; qp < SCALING_LIST_REM_NUM; qp++)
        {
          if(m_quantCoef[sizeIdX][sizeIdY][listId][qp])
          {
            delete [] m_quantCoef[sizeIdX][sizeIdY][listId][qp];
          }
          if(m_dequantCoef[sizeIdX][sizeIdY][listId][qp])
          {
            delete [] m_dequantCoef[sizeIdX][sizeIdY][listId][qp];
          }
        }
      }
    }
  }
}

void Quant::quant(TransformUnit& tu, const ComponentID compID, const CCoeffBuf& pSrc, TCoeff &uiAbsSum, const QpParam& cQP, const Ctx& ctx)
{
  const SPS &sps            = *tu.cs->sps;
  const CompArea& rect      = tu.blocks[compID];
  const uint32_t uiWidth    = rect.width;
  const uint32_t uiHeight   = rect.height;
  const int channelBitDepth = sps.bitDepths[toChannelType(compID)];

  const CCoeffBuf&  piCoef  = pSrc;
        CoeffSigBuf piQCoef = tu.getCoeffs(compID);

  const bool useTransformSkip = tu.mtsIdx[compID] == MTS_SKIP;
  const int  maxLog2TrDynamicRange = sps.getMaxLog2TrDynamicRange();

  {
    CoeffCodingContext cctx(tu, compID, tu.cs->slice->signDataHidingEnabled);

    const TCoeff entropyCodingMinimum = -(1 << maxLog2TrDynamicRange);
    const TCoeff entropyCodingMaximum =  (1 << maxLog2TrDynamicRange) - 1;

    TCoeff deltaU[MAX_TB_SIZEY * MAX_TB_SIZEY];
    int scalingListType           = getScalingListType(tu.cu->predMode, compID);
    CHECK(scalingListType >= SCALING_LIST_NUM, "Invalid scaling list");
    const uint32_t uiLog2TrWidth  = Log2(uiWidth);
    const uint32_t uiLog2TrHeight = Log2(uiHeight);
    int *piQuantCoeff             = getQuantCoeff(scalingListType, cQP.rem(useTransformSkip), uiLog2TrWidth, uiLog2TrHeight);

    const bool isLfnstApplied     = tu.cu->lfnstIdx > 0 && (CU::isSepTree(*tu.cu) ? true : isLuma(compID));
    const bool enableScalingLists = getUseScalingList(uiWidth, uiHeight, useTransformSkip, isLfnstApplied);

    // for blocks that where width*height != 4^N, the effective scaling applied during transformation cannot be
    // compensated by a bit-shift (the quantised result will be sqrt(2) * larger than required).
    // The quantScale table and shift is used to compensate for this.
    const bool needSqrtAdjustment= TU::needsSqrt2Scale( tu, compID );
    const int defaultQuantisationCoefficient    = g_quantScales[needSqrtAdjustment?1:0][cQP.rem(useTransformSkip)];
    const int iTransformShift = getTransformShift(channelBitDepth, rect.size(), maxLog2TrDynamicRange) + ( needSqrtAdjustment?-1:0);

    const int iQBits = QUANT_SHIFT + cQP.per(useTransformSkip) + (useTransformSkip ? 0 : iTransformShift);
    // QBits will be OK for any internal bit depth as the reduction in transform shift is balanced by an increase in Qp_per due to QpBDOffset

    const int64_t iAdd = int64_t(tu.cs->slice->isIRAP() ? 171 : 85) << int64_t(iQBits - 9);
    const int qBits8 = iQBits - 8;

    int lastScanPos = -1;

    if (!enableScalingLists)
      xQuant(tu,compID,piCoef,piQCoef,uiAbsSum,lastScanPos,deltaU,defaultQuantisationCoefficient,iQBits,iAdd,entropyCodingMinimum,entropyCodingMaximum,cctx.signHiding(),m_thrVal);
    else
    {
      const uint32_t lfnstIdx = tu.cu->lfnstIdx;
      const int maxNumberOfCoeffs = lfnstIdx > 0 ? ( ( ( uiWidth == 4 && uiHeight == 4 ) || ( uiWidth == 8 && uiHeight == 8 ) ) ? 8 : 16 ) : piQCoef.area();

      piQCoef.memset( 0 );
      for (int uiScanPos = 0; uiScanPos < maxNumberOfCoeffs; uiScanPos++ )
      {
        const int uiBlockPos  = cctx.blockPos( uiScanPos );
        const TCoeff iLevel   = piCoef.buf[uiBlockPos];
        const TCoeff iSign    = (iLevel < 0 ? -1: 1);

        const int64_t  tmpLevel = (int64_t)abs(iLevel) * (enableScalingLists ? piQuantCoeff[uiBlockPos] : defaultQuantisationCoefficient);
        const TCoeff quantisedMagnitude = TCoeff((tmpLevel + iAdd ) >> iQBits);
        deltaU[uiBlockPos] = (TCoeff)((tmpLevel - ((int64_t)quantisedMagnitude<<iQBits) )>> qBits8);

        uiAbsSum += quantisedMagnitude;
        const TCoeff quantisedCoefficient = quantisedMagnitude * iSign;

        piQCoef.buf[uiBlockPos] = Clip3<TCoeff>( entropyCodingMinimum, entropyCodingMaximum, quantisedCoefficient );
      } // for n
    }
    if (tu.cu->bdpcmM[toChannelType(compID)])
    {
      CoeffSigBuf coeffs = tu.getCoeffs( compID );
      xFwdResDPCM( coeffs.buf, coeffs.stride, uiWidth, uiHeight, tu.cu->bdpcmM[toChannelType( compID )] != 1 );
    }

    if( uiAbsSum )
    {
      for( int scanPos = lastScanPos; scanPos >= 0; scanPos-- )
      {
        unsigned blkPos = cctx.blockPos( scanPos );
        if( piQCoef.buf[blkPos] )
        {
          lastScanPos = scanPos;
          break;
        }
      }

      if( cctx.signHiding() )
      {
        if( uiAbsSum >= 2 ) //this prevents TUs with only one coefficient of value 1 from being tested
        {
          xSignBitHidingHDQ( piQCoef.buf, piCoef.buf, deltaU, cctx, lastScanPos, maxLog2TrDynamicRange );
        }
      }
    }

    tu.lastPos[compID] = lastScanPos;
  } //if RDOQ
  //return;
}

bool Quant::xNeedRDOQ(TransformUnit& tu, const ComponentID compID, const CCoeffBuf& pSrc, const QpParam& cQP)
{
  const SPS &sps            = *tu.cs->sps;
  const CompArea& rect      = tu.blocks[compID];
  const uint32_t uiWidth    = rect.width;
  const uint32_t uiHeight   = rect.height;
  const uint32_t efHeight   = std::min<unsigned>( uiHeight, JVET_C0024_ZERO_OUT_TH );
  const uint32_t efArea     = uiWidth * efHeight;
  const int channelBitDepth = sps.bitDepths[toChannelType(compID)];
  const CCoeffBuf piCoef    = pSrc;

  const bool useTransformSkip      = tu.mtsIdx[compID] == MTS_SKIP;
  const int  maxLog2TrDynamicRange = sps.getMaxLog2TrDynamicRange();

  const int scalingListType     = getScalingListType( tu.cu->predMode, compID );
  CHECK( scalingListType >= SCALING_LIST_NUM, "Invalid scaling list" );

  const bool        isDq        = tu.cs->slice->depQuantEnabled && !useTransformSkip;
  const int         qpDQ        = isDq ? cQP.Qp( false ) + 1 : cQP.Qp( useTransformSkip );
  const int         qpPer       = isDq ? qpDQ / 6 : cQP.per( useTransformSkip );
  const int         qpRem       = isDq ? qpDQ - 6 * qpPer : cQP.rem( useTransformSkip );

  const uint32_t uiLog2TrWidth  = Log2( uiWidth );
  const uint32_t uiLog2TrHeight = Log2( uiHeight );
  int *piQuantCoeff             = getQuantCoeff( scalingListType, qpRem, uiLog2TrWidth, uiLog2TrHeight );

  const bool isLfnstApplied     = tu.cu->lfnstIdx > 0 && ( CU::isSepTree( *tu.cu ) ? true : isLuma( compID ) );
  const bool enableScalingLists = getUseScalingList( uiWidth, uiHeight, ( useTransformSkip != 0 ), isLfnstApplied );

  const bool needSqrtAdjustment = TU::needsSqrt2Scale( tu, compID );
  const int defaultQuantisationCoefficient
                                = g_quantScales[needSqrtAdjustment?1:0][qpRem];
  const int iTransformShift     = getTransformShift( channelBitDepth, rect.size(), maxLog2TrDynamicRange ) + ( needSqrtAdjustment ? -1 : 0 );


  const int iQBits              = QUANT_SHIFT + qpPer + iTransformShift;

  // QBits will be OK for any internal bit depth as the reduction in transform shift is balanced by an increase in Qp_per due to QpBDOffset
  // iAdd is different from the iAdd used in normal quantization
  const int64_t iAdd = int64_t( compID == COMP_Y ? 171 : 256 ) << ( iQBits - 9 );

  if( !enableScalingLists )
    return xNeedRdoq( piCoef.buf, efArea, defaultQuantisationCoefficient, iAdd, iQBits );

  for( int uiBlockPos = 0; uiBlockPos < efArea; uiBlockPos++ )
  {
    const TCoeff   iLevel           = piCoef.buf[uiBlockPos];
    const int64_t  tmpLevel         = ( int64_t ) std::abs( iLevel ) * piQuantCoeff[uiBlockPos];
    const TCoeff quantisedMagnitude = TCoeff( ( tmpLevel + iAdd ) >> iQBits );

    if( quantisedMagnitude != 0 )
    {
      return true;
    }
  } // for n
  return false;
}

} // namespace vvenc

//! \}


Coverage Report

Created: 2026-10-28 06:18