UnitTools.cpp

  mvLT = puNeighbour->mvAffi[eRefPicList][0];
  mvRT = puNeighbour->mvAffi[eRefPicList][1];
  mvLB = puNeighbour->mvAffi[eRefPicList][2];

  bool isTopCtuBoundary = false;
  if ( (posNeiY + neiH) % pu.cs->sps->getCTUSize() == 0 && (posNeiY + neiH) == posCurY )
  {
    // use bottom-left and bottom-right sub-block MVs for inheritance
    const Position posRB = puNeighbour->Y().bottomRight();
    const Position posLB = puNeighbour->Y().bottomLeft();
    mvLT = puNeighbour->getMotionInfo( posLB ).mv[eRefPicList];
    mvRT = puNeighbour->getMotionInfo( posRB ).mv[eRefPicList];
    posNeiY += neiH;
    isTopCtuBoundary = true;
  }

  int shift = MAX_CU_DEPTH;
  int iDMvHorX, iDMvHorY, iDMvVerX, iDMvVerY;

  iDMvHorX = (mvRT - mvLT).getHor() << (shift - floorLog2(neiW));
  iDMvHorY = (mvRT - mvLT).getVer() << (shift - floorLog2(neiW));
  if ( puNeighbour->cu->affineType == AFFINEMODEL_6PARAM && !isTopCtuBoundary )
  {
    iDMvVerX = (mvLB - mvLT).getHor() << (shift - floorLog2(neiH));
    iDMvVerY = (mvLB - mvLT).getVer() << (shift - floorLog2(neiH));
  }
  else
  {
    iDMvVerX = -iDMvHorY;
    iDMvVerY = iDMvHorX;
  }

  int iMvScaleHor = mvLT.getHor() << shift;
  int iMvScaleVer = mvLT.getVer() << shift;
  int horTmp, verTmp;

  // v0
  horTmp = iMvScaleHor + iDMvHorX * (posCurX - posNeiX) + iDMvVerX * (posCurY - posNeiY);
  verTmp = iMvScaleVer + iDMvHorY * (posCurX - posNeiX) + iDMvVerY * (posCurY - posNeiY);
  roundAffineMv( horTmp, verTmp, shift );
  rcMv[0].hor = horTmp;
  rcMv[0].ver = verTmp;
  rcMv[0].clipToStorageBitDepth();

  // v1
  horTmp = iMvScaleHor + iDMvHorX * (posCurX + curW - posNeiX) + iDMvVerX * (posCurY - posNeiY);
  verTmp = iMvScaleVer + iDMvHorY * (posCurX + curW - posNeiX) + iDMvVerY * (posCurY - posNeiY);
  roundAffineMv( horTmp, verTmp, shift );
  rcMv[1].hor = horTmp;
  rcMv[1].ver = verTmp;
  rcMv[1].clipToStorageBitDepth();

  // v2
  if ( pu.cu->affineType == AFFINEMODEL_6PARAM )
  {
    horTmp = iMvScaleHor + iDMvHorX * (posCurX - posNeiX) + iDMvVerX * (posCurY + curH - posNeiY);
    verTmp = iMvScaleVer + iDMvHorY * (posCurX - posNeiX) + iDMvVerY * (posCurY + curH - posNeiY);
    roundAffineMv( horTmp, verTmp, shift );
    rcMv[2].hor = horTmp;
    rcMv[2].ver = verTmp;
    rcMv[2].clipToStorageBitDepth();
  }
}


void PU::fillAffineMvpCand(PredictionUnit &pu, const RefPicList &eRefPicList, const int &refIdx, AffineAMVPInfo &affiAMVPInfo)
{
  affiAMVPInfo.numCand = 0;

  if (refIdx < 0)
  {
    return;
  }


  // insert inherited affine candidates
  Mv outputAffineMv[3];
  Position posLT = pu.Y().topLeft();
  Position posRT = pu.Y().topRight();
  Position posLB = pu.Y().bottomLeft();

  // check left neighbor
  if ( !addAffineMVPCandUnscaled( pu, eRefPicList, refIdx, posLB, MD_BELOW_LEFT, affiAMVPInfo ) )
  {
    addAffineMVPCandUnscaled( pu, eRefPicList, refIdx, posLB, MD_LEFT, affiAMVPInfo );
  }

  // check above neighbor
  if ( !addAffineMVPCandUnscaled( pu, eRefPicList, refIdx, posRT, MD_ABOVE_RIGHT, affiAMVPInfo ) )
  {
    if ( !addAffineMVPCandUnscaled( pu, eRefPicList, refIdx, posRT, MD_ABOVE, affiAMVPInfo ) )
    {
      addAffineMVPCandUnscaled( pu, eRefPicList, refIdx, posLT, MD_ABOVE_LEFT, affiAMVPInfo );
    }
  }

  if ( affiAMVPInfo.numCand >= AMVP_MAX_NUM_CANDS )
  {
    for (int i = 0; i < affiAMVPInfo.numCand; i++)
    {
      affiAMVPInfo.mvCandLT[i].roundAffinePrecInternal2Amvr(pu.cu->imv);
      affiAMVPInfo.mvCandRT[i].roundAffinePrecInternal2Amvr(pu.cu->imv);
      affiAMVPInfo.mvCandLB[i].roundAffinePrecInternal2Amvr(pu.cu->imv);
    }
    return;
  }

  // insert constructed affine candidates
  int cornerMVPattern = 0;

  //-------------------  V0 (START) -------------------//
  AMVPInfo amvpInfo0;
  amvpInfo0.numCand = 0;

  // A->C: Above Left, Above, Left
  addMVPCandUnscaled( pu, eRefPicList, refIdx, posLT, MD_ABOVE_LEFT, amvpInfo0 );
  if ( amvpInfo0.numCand < 1 )
  {
    addMVPCandUnscaled( pu, eRefPicList, refIdx, posLT, MD_ABOVE, amvpInfo0 );
  }
  if ( amvpInfo0.numCand < 1 )
  {
    addMVPCandUnscaled( pu, eRefPicList, refIdx, posLT, MD_LEFT, amvpInfo0 );
  }
  cornerMVPattern = cornerMVPattern | amvpInfo0.numCand;

  //-------------------  V1 (START) -------------------//
  AMVPInfo amvpInfo1;
  amvpInfo1.numCand = 0;

  // D->E: Above, Above Right
  addMVPCandUnscaled( pu, eRefPicList, refIdx, posRT, MD_ABOVE, amvpInfo1 );
  if ( amvpInfo1.numCand < 1 )
  {
    addMVPCandUnscaled( pu, eRefPicList, refIdx, posRT, MD_ABOVE_RIGHT, amvpInfo1 );
  }
  cornerMVPattern = cornerMVPattern | (amvpInfo1.numCand << 1);

  //-------------------  V2 (START) -------------------//
  AMVPInfo amvpInfo2;
  amvpInfo2.numCand = 0;

  // F->G: Left, Below Left
  addMVPCandUnscaled( pu, eRefPicList, refIdx, posLB, MD_LEFT, amvpInfo2 );
  if ( amvpInfo2.numCand < 1 )
  {
    addMVPCandUnscaled( pu, eRefPicList, refIdx, posLB, MD_BELOW_LEFT, amvpInfo2 );
  }
  cornerMVPattern = cornerMVPattern | (amvpInfo2.numCand << 2);

  outputAffineMv[0] = amvpInfo0.mvCand[0];
  outputAffineMv[1] = amvpInfo1.mvCand[0];
  outputAffineMv[2] = amvpInfo2.mvCand[0];

  outputAffineMv[0].roundAffinePrecInternal2Amvr(pu.cu->imv);
  outputAffineMv[1].roundAffinePrecInternal2Amvr(pu.cu->imv);
  outputAffineMv[2].roundAffinePrecInternal2Amvr(pu.cu->imv);

  if ( cornerMVPattern == 7 || (cornerMVPattern == 3 && pu.cu->affineType == AFFINEMODEL_4PARAM) )
  {
    affiAMVPInfo.mvCandLT[affiAMVPInfo.numCand] = outputAffineMv[0];
    affiAMVPInfo.mvCandRT[affiAMVPInfo.numCand] = outputAffineMv[1];
    affiAMVPInfo.mvCandLB[affiAMVPInfo.numCand] = outputAffineMv[2];
    affiAMVPInfo.numCand++;
  }


  if ( affiAMVPInfo.numCand < 2 )
  {
    // check corner MVs
    for ( int i = 2; i >= 0 && affiAMVPInfo.numCand < AMVP_MAX_NUM_CANDS; i-- )
    {
      if ( cornerMVPattern & (1 << i) ) // MV i exist
      {
        affiAMVPInfo.mvCandLT[affiAMVPInfo.numCand] = outputAffineMv[i];
        affiAMVPInfo.mvCandRT[affiAMVPInfo.numCand] = outputAffineMv[i];
        affiAMVPInfo.mvCandLB[affiAMVPInfo.numCand] = outputAffineMv[i];
        affiAMVPInfo.numCand++;
      }
    }

    // Get Temporal Motion Predictor
#if JVET_P1006_PICTURE_HEADER
    if ( affiAMVPInfo.numCand < 2 && pu.cs->picHeader->getEnableTMVPFlag() )
#else
    if ( affiAMVPInfo.numCand < 2 && pu.cs->slice->getEnableTMVPFlag() )
#endif
    {
      const int refIdxCol = refIdx;

      Position posRB = pu.Y().bottomRight().offset( -3, -3 );

      const PreCalcValues& pcv = *pu.cs->pcv;

      Position posC0;
      bool C0Avail = false;
      Position posC1 = pu.Y().center();
      Mv cColMv;
      if ( ((posRB.x + pcv.minCUWidth) < pcv.lumaWidth) && ((posRB.y + pcv.minCUHeight) < pcv.lumaHeight) )
      {
        int posYInCtu = posRB.y & pcv.maxCUHeightMask;
        if (posYInCtu + 4 < pcv.maxCUHeight)
        {
          posC0 = posRB.offset(4, 4);
          C0Avail = true;
        }
      }
      if ( ( C0Avail && getColocatedMVP( pu, eRefPicList, posC0, cColMv, refIdxCol, false ) ) || getColocatedMVP( pu, eRefPicList, posC1, cColMv, refIdxCol, false ) )
      {
        cColMv.roundAffinePrecInternal2Amvr(pu.cu->imv);
        affiAMVPInfo.mvCandLT[affiAMVPInfo.numCand] = cColMv;
        affiAMVPInfo.mvCandRT[affiAMVPInfo.numCand] = cColMv;
        affiAMVPInfo.mvCandLB[affiAMVPInfo.numCand] = cColMv;
        affiAMVPInfo.numCand++;
      }
    }

    if ( affiAMVPInfo.numCand < 2 )
    {
      // add zero MV
      for ( int i = affiAMVPInfo.numCand; i < AMVP_MAX_NUM_CANDS; i++ )
      {
        affiAMVPInfo.mvCandLT[affiAMVPInfo.numCand].setZero();
        affiAMVPInfo.mvCandRT[affiAMVPInfo.numCand].setZero();
        affiAMVPInfo.mvCandLB[affiAMVPInfo.numCand].setZero();
        affiAMVPInfo.numCand++;
      }
    }
  }

  for (int i = 0; i < affiAMVPInfo.numCand; i++)
  {
    affiAMVPInfo.mvCandLT[i].roundAffinePrecInternal2Amvr(pu.cu->imv);
    affiAMVPInfo.mvCandRT[i].roundAffinePrecInternal2Amvr(pu.cu->imv);
    affiAMVPInfo.mvCandLB[i].roundAffinePrecInternal2Amvr(pu.cu->imv);
  }
}

bool PU::addMVPCandUnscaled( const PredictionUnit &pu, const RefPicList &eRefPicList, const int &iRefIdx, const Position &pos, const MvpDir &eDir, AMVPInfo &info )
{
        CodingStructure &cs    = *pu.cs;
  const PredictionUnit *neibPU = NULL;
        Position neibPos;

  switch (eDir)
  {
  case MD_LEFT:
    neibPos = pos.offset( -1,  0 );
    break;
  case MD_ABOVE:
    neibPos = pos.offset(  0, -1 );
    break;
  case MD_ABOVE_RIGHT:
    neibPos = pos.offset(  1, -1 );
    break;
  case MD_BELOW_LEFT:
    neibPos = pos.offset( -1,  1 );
    break;
  case MD_ABOVE_LEFT:
    neibPos = pos.offset( -1, -1 );
    break;
  default:
    break;
  }

  neibPU = cs.getPURestricted( neibPos, pu, pu.chType );

  if( neibPU == NULL || !CU::isInter( *neibPU->cu ) )
  {
    return false;
  }

  const MotionInfo& neibMi        = neibPU->getMotionInfo( neibPos );

  const int        currRefPOC     = cs.slice->getRefPic( eRefPicList, iRefIdx )->getPOC();
  const RefPicList eRefPicList2nd = ( eRefPicList == REF_PIC_LIST_0 ) ? REF_PIC_LIST_1 : REF_PIC_LIST_0;

  for( int predictorSource = 0; predictorSource < 2; predictorSource++ ) // examine the indicated reference picture list, then if not available, examine the other list.
  {
    const RefPicList eRefPicListIndex = ( predictorSource == 0 ) ? eRefPicList : eRefPicList2nd;
    const int        neibRefIdx       = neibMi.refIdx[eRefPicListIndex];

    if( neibRefIdx >= 0 && currRefPOC == cs.slice->getRefPOC( eRefPicListIndex, neibRefIdx ) )
    {
      info.mvCand[info.numCand++] = neibMi.mv[eRefPicListIndex];
      return true;
    }
  }

  return false;
}


void PU::addAMVPHMVPCand(const PredictionUnit &pu, const RefPicList eRefPicList, const int currRefPOC, AMVPInfo &info)
{
  const Slice &slice = *(*pu.cs).slice;

  MotionInfo neibMi;
  auto &lut = CU::isIBC(*pu.cu) ? pu.cs->motionLut.lutIbc : pu.cs->motionLut.lut;
  int num_avai_candInLUT = (int) lut.size();
  int num_allowedCand = std::min(MAX_NUM_HMVP_AVMPCANDS, num_avai_candInLUT);
  const RefPicList eRefPicList2nd = (eRefPicList == REF_PIC_LIST_0) ? REF_PIC_LIST_1 : REF_PIC_LIST_0;

  for (int mrgIdx = 1; mrgIdx <= num_allowedCand; mrgIdx++)
  {
    if (info.numCand >= AMVP_MAX_NUM_CANDS)
    {
      return;
    }
    neibMi = lut[mrgIdx - 1];

    for (int predictorSource = 0; predictorSource < 2; predictorSource++)
    {
      const RefPicList eRefPicListIndex = (predictorSource == 0) ? eRefPicList : eRefPicList2nd;
      const int        neibRefIdx = neibMi.refIdx[eRefPicListIndex];

      if (neibRefIdx >= 0 && (CU::isIBC(*pu.cu) || (currRefPOC == slice.getRefPOC(eRefPicListIndex, neibRefIdx))))
      {
        Mv pmv = neibMi.mv[eRefPicListIndex];
        pmv.roundTransPrecInternal2Amvr(pu.cu->imv);

        info.mvCand[info.numCand++] = pmv;
        if (info.numCand >= AMVP_MAX_NUM_CANDS)
        {
          return;
        }
      }
    }
  }
}

bool PU::isBipredRestriction(const PredictionUnit &pu)
{
  if(pu.cu->lumaSize().width == 4 && pu.cu->lumaSize().height ==4 )
  {
    return true;
  }
  /* disable bi-prediction for 4x8/8x4 */
  if ( pu.cu->lumaSize().width + pu.cu->lumaSize().height == 12 )
  {
    return true;
  }
  return false;
}

void PU::getAffineControlPointCand(const PredictionUnit &pu, MotionInfo mi[4], bool isAvailable[4], int verIdx[4], int8_t gbiIdx, int modelIdx, int verNum, AffineMergeCtx& affMrgType)
{
  int cuW = pu.Y().width;
  int cuH = pu.Y().height;
  int vx, vy;
  int shift = MAX_CU_DEPTH;
  int shiftHtoW = shift + floorLog2(cuW) - floorLog2(cuH);

  // motion info
  Mv cMv[2][4];
  int refIdx[2] = { -1, -1 };
  int dir = 0;
  EAffineModel curType = (verNum == 2) ? AFFINEMODEL_4PARAM : AFFINEMODEL_6PARAM;

  if ( verNum == 2 )
  {
    int idx0 = verIdx[0], idx1 = verIdx[1];
    if ( !isAvailable[idx0] || !isAvailable[idx1] )
    {
      return;
    }

    for ( int l = 0; l < 2; l++ )
    {
      if ( mi[idx0].refIdx[l] >= 0 && mi[idx1].refIdx[l] >= 0 )
      {
        // check same refidx and different mv
        if ( mi[idx0].refIdx[l] == mi[idx1].refIdx[l])
        {
          dir |= (l + 1);
          refIdx[l] = mi[idx0].refIdx[l];
        }
      }
    }

  }
  else if ( verNum == 3 )
  {
    int idx0 = verIdx[0], idx1 = verIdx[1], idx2 = verIdx[2];
    if ( !isAvailable[idx0] || !isAvailable[idx1] || !isAvailable[idx2] )
    {
      return;
    }

    for ( int l = 0; l < 2; l++ )
    {
      if ( mi[idx0].refIdx[l] >= 0 && mi[idx1].refIdx[l] >= 0 && mi[idx2].refIdx[l] >= 0 )
      {
        // check same refidx and different mv
        if ( mi[idx0].refIdx[l] == mi[idx1].refIdx[l] && mi[idx0].refIdx[l] == mi[idx2].refIdx[l])
        {
          dir |= (l + 1);
          refIdx[l] = mi[idx0].refIdx[l];
        }
      }
    }

  }

  if ( dir == 0 )
  {
    return;
  }


  for ( int l = 0; l < 2; l++ )
  {
    if ( dir & (l + 1) )
    {
      for ( int i = 0; i < verNum; i++ )
      {
        cMv[l][verIdx[i]] = mi[verIdx[i]].mv[l];
      }

      // convert to LT, RT[, [LB]]
      switch ( modelIdx )
      {
      case 0: // 0 : LT, RT, LB
        break;

      case 1: // 1 : LT, RT, RB
        cMv[l][2].hor = cMv[l][3].hor + cMv[l][0].hor - cMv[l][1].hor;
        cMv[l][2].ver = cMv[l][3].ver + cMv[l][0].ver - cMv[l][1].ver;
        cMv[l][2].clipToStorageBitDepth();
        break;

      case 2: // 2 : LT, LB, RB
        cMv[l][1].hor = cMv[l][3].hor + cMv[l][0].hor - cMv[l][2].hor;
        cMv[l][1].ver = cMv[l][3].ver + cMv[l][0].ver - cMv[l][2].ver;
        cMv[l][1].clipToStorageBitDepth();
        break;

      case 3: // 3 : RT, LB, RB
        cMv[l][0].hor = cMv[l][1].hor + cMv[l][2].hor - cMv[l][3].hor;
        cMv[l][0].ver = cMv[l][1].ver + cMv[l][2].ver - cMv[l][3].ver;
        cMv[l][0].clipToStorageBitDepth();
        break;

      case 4: // 4 : LT, RT
        break;

      case 5: // 5 : LT, LB
        vx = (cMv[l][0].hor << shift) + ((cMv[l][2].ver - cMv[l][0].ver) << shiftHtoW);
        vy = (cMv[l][0].ver << shift) - ((cMv[l][2].hor - cMv[l][0].hor) << shiftHtoW);
        roundAffineMv( vx, vy, shift );
        cMv[l][1].set( vx, vy );
        cMv[l][1].clipToStorageBitDepth();
        break;

      default:
        CHECK( 1, "Invalid model index!\n" );
        break;
      }
    }
    else
    {
      for ( int i = 0; i < 4; i++ )
      {
        cMv[l][i].hor = 0;
        cMv[l][i].ver = 0;
      }
    }
  }

  for ( int i = 0; i < 3; i++ )
  {
    affMrgType.mvFieldNeighbours[(affMrgType.numValidMergeCand << 1) + 0][i].mv = cMv[0][i];
    affMrgType.mvFieldNeighbours[(affMrgType.numValidMergeCand << 1) + 0][i].refIdx = refIdx[0];

    affMrgType.mvFieldNeighbours[(affMrgType.numValidMergeCand << 1) + 1][i].mv = cMv[1][i];
    affMrgType.mvFieldNeighbours[(affMrgType.numValidMergeCand << 1) + 1][i].refIdx = refIdx[1];
  }
  affMrgType.interDirNeighbours[affMrgType.numValidMergeCand] = dir;
  affMrgType.affineType[affMrgType.numValidMergeCand] = curType;
  affMrgType.GBiIdx[affMrgType.numValidMergeCand] = (dir == 3) ? gbiIdx : GBI_DEFAULT;
  affMrgType.numValidMergeCand++;


  return;
}

const int getAvailableAffineNeighboursForLeftPredictor( const PredictionUnit &pu, const PredictionUnit* npu[] )
{
  const Position posLB = pu.Y().bottomLeft();
  int num = 0;

  const PredictionUnit *puLeftBottom = pu.cs->getPURestricted( posLB.offset( -1, 1 ), pu, pu.chType );
  if ( puLeftBottom && puLeftBottom->cu->affine
    && puLeftBottom->mergeType == MRG_TYPE_DEFAULT_N
    )
  {
    npu[num++] = puLeftBottom;
    return num;
  }

  const PredictionUnit* puLeft = pu.cs->getPURestricted( posLB.offset( -1, 0 ), pu, pu.chType );
  if ( puLeft && puLeft->cu->affine
    && puLeft->mergeType == MRG_TYPE_DEFAULT_N
    )
  {
    npu[num++] = puLeft;
    return num;
  }

  return num;
}

const int getAvailableAffineNeighboursForAbovePredictor( const PredictionUnit &pu, const PredictionUnit* npu[], int numAffNeighLeft )
{
  const Position posLT = pu.Y().topLeft();
  const Position posRT = pu.Y().topRight();
  int num = numAffNeighLeft;

  const PredictionUnit* puAboveRight = pu.cs->getPURestricted( posRT.offset( 1, -1 ), pu, pu.chType );
  if ( puAboveRight && puAboveRight->cu->affine
    && puAboveRight->mergeType == MRG_TYPE_DEFAULT_N
    )
  {
    npu[num++] = puAboveRight;
    return num;
  }

  const PredictionUnit* puAbove = pu.cs->getPURestricted( posRT.offset( 0, -1 ), pu, pu.chType );
  if ( puAbove && puAbove->cu->affine
    && puAbove->mergeType == MRG_TYPE_DEFAULT_N
    )
  {
    npu[num++] = puAbove;
    return num;
  }

  const PredictionUnit *puAboveLeft = pu.cs->getPURestricted( posLT.offset( -1, -1 ), pu, pu.chType );
  if ( puAboveLeft && puAboveLeft->cu->affine
    && puAboveLeft->mergeType == MRG_TYPE_DEFAULT_N
    )
  {
    npu[num++] = puAboveLeft;
    return num;
  }

  return num;
}

void PU::getAffineMergeCand( const PredictionUnit &pu, AffineMergeCtx& affMrgCtx, const int mrgCandIdx )
{
  const CodingStructure &cs = *pu.cs;
  const Slice &slice = *pu.cs->slice;
#if JVET_P1006_PICTURE_HEADER
  const uint32_t maxNumAffineMergeCand = slice.getPicHeader()->getMaxNumAffineMergeCand();
#else
  const uint32_t maxNumAffineMergeCand = slice.getMaxNumAffineMergeCand();
#endif

  for ( int i = 0; i < maxNumAffineMergeCand; i++ )
  {
    for ( int mvNum = 0; mvNum < 3; mvNum++ )
    {
      affMrgCtx.mvFieldNeighbours[(i << 1) + 0][mvNum].setMvField( Mv(), -1 );
      affMrgCtx.mvFieldNeighbours[(i << 1) + 1][mvNum].setMvField( Mv(), -1 );
    }
    affMrgCtx.interDirNeighbours[i] = 0;
    affMrgCtx.affineType[i] = AFFINEMODEL_4PARAM;
    affMrgCtx.mergeType[i] = MRG_TYPE_DEFAULT_N;
    affMrgCtx.GBiIdx[i] = GBI_DEFAULT;
  }

  affMrgCtx.numValidMergeCand = 0;
  affMrgCtx.maxNumMergeCand = maxNumAffineMergeCand;

  bool enableSubPuMvp = slice.getSPS()->getSBTMVPEnabledFlag() && !(slice.getPOC() == slice.getRefPic(REF_PIC_LIST_0, 0)->getPOC() && slice.isIRAP());
  bool isAvailableSubPu = false;
#if JVET_P1006_PICTURE_HEADER
  if ( enableSubPuMvp && slice.getPicHeader()->getEnableTMVPFlag() )
#else
  if ( enableSubPuMvp && slice.getEnableTMVPFlag() )
#endif
  {
    MergeCtx mrgCtx = *affMrgCtx.mrgCtx;
    bool tmpLICFlag = false;

    CHECK( mrgCtx.subPuMvpMiBuf.area() == 0 || !mrgCtx.subPuMvpMiBuf.buf, "Buffer not initialized" );
    mrgCtx.subPuMvpMiBuf.fill( MotionInfo() );

    int pos = 0;
    // Get spatial MV
    const Position posCurLB = pu.Y().bottomLeft();
    MotionInfo miLeft;

    //left
    const PredictionUnit* puLeft = cs.getPURestricted( posCurLB.offset( -1, 0 ), pu, pu.chType );
    const bool isAvailableA1 = puLeft && isDiffMER( pu, *puLeft ) && pu.cu != puLeft->cu && CU::isInter( *puLeft->cu );
    if ( isAvailableA1 )
    {
      miLeft = puLeft->getMotionInfo( posCurLB.offset( -1, 0 ) );
      // get Inter Dir
      mrgCtx.interDirNeighbours[pos] = miLeft.interDir;

      // get Mv from Left
      mrgCtx.mvFieldNeighbours[pos << 1].setMvField( miLeft.mv[0], miLeft.refIdx[0] );

      if ( slice.isInterB() )
      {
        mrgCtx.mvFieldNeighbours[(pos << 1) + 1].setMvField( miLeft.mv[1], miLeft.refIdx[1] );
      }
      pos++;
    }

    mrgCtx.numValidMergeCand = pos;

    isAvailableSubPu = getInterMergeSubPuMvpCand( pu, mrgCtx, tmpLICFlag, pos
      , 0
    );
    if ( isAvailableSubPu )
    {
      for ( int mvNum = 0; mvNum < 3; mvNum++ )
      {
        affMrgCtx.mvFieldNeighbours[(affMrgCtx.numValidMergeCand << 1) + 0][mvNum].setMvField( mrgCtx.mvFieldNeighbours[(pos << 1) + 0].mv, mrgCtx.mvFieldNeighbours[(pos << 1) + 0].refIdx );
        affMrgCtx.mvFieldNeighbours[(affMrgCtx.numValidMergeCand << 1) + 1][mvNum].setMvField( mrgCtx.mvFieldNeighbours[(pos << 1) + 1].mv, mrgCtx.mvFieldNeighbours[(pos << 1) + 1].refIdx );
      }
      affMrgCtx.interDirNeighbours[affMrgCtx.numValidMergeCand] = mrgCtx.interDirNeighbours[pos];

      affMrgCtx.affineType[affMrgCtx.numValidMergeCand] = AFFINE_MODEL_NUM;
      affMrgCtx.mergeType[affMrgCtx.numValidMergeCand] = MRG_TYPE_SUBPU_ATMVP;
      if ( affMrgCtx.numValidMergeCand == mrgCandIdx )
      {
        return;
      }

      affMrgCtx.numValidMergeCand++;

      // early termination
      if ( affMrgCtx.numValidMergeCand == maxNumAffineMergeCand )
      {
        return;
      }
    }
  }

  if ( slice.getSPS()->getUseAffine() )
  {
    ///> Start: inherited affine candidates
    const PredictionUnit* npu[5];
    int numAffNeighLeft = getAvailableAffineNeighboursForLeftPredictor( pu, npu );
    int numAffNeigh = getAvailableAffineNeighboursForAbovePredictor( pu, npu, numAffNeighLeft );
    for ( int idx = 0; idx < numAffNeigh; idx++ )
    {
      // derive Mv from Neigh affine PU
      Mv cMv[2][3];
      const PredictionUnit* puNeigh = npu[idx];
      pu.cu->affineType = puNeigh->cu->affineType;
      if ( puNeigh->interDir != 2 )
      {
        xInheritedAffineMv( pu, puNeigh, REF_PIC_LIST_0, cMv[0] );
      }
      if ( slice.isInterB() )
      {
        if ( puNeigh->interDir != 1 )
        {
          xInheritedAffineMv( pu, puNeigh, REF_PIC_LIST_1, cMv[1] );
        }
      }

      for ( int mvNum = 0; mvNum < 3; mvNum++ )
      {
        affMrgCtx.mvFieldNeighbours[(affMrgCtx.numValidMergeCand << 1) + 0][mvNum].setMvField( cMv[0][mvNum], puNeigh->refIdx[0] );
        affMrgCtx.mvFieldNeighbours[(affMrgCtx.numValidMergeCand << 1) + 1][mvNum].setMvField( cMv[1][mvNum], puNeigh->refIdx[1] );
      }
      affMrgCtx.interDirNeighbours[affMrgCtx.numValidMergeCand] = puNeigh->interDir;
      affMrgCtx.affineType[affMrgCtx.numValidMergeCand] = (EAffineModel)(puNeigh->cu->affineType);
      affMrgCtx.GBiIdx[affMrgCtx.numValidMergeCand] = puNeigh->cu->GBiIdx;

      if ( affMrgCtx.numValidMergeCand == mrgCandIdx )
      {
        return;
      }

      // early termination
      affMrgCtx.numValidMergeCand++;
      if ( affMrgCtx.numValidMergeCand == maxNumAffineMergeCand )
      {
        return;
      }
    }
    ///> End: inherited affine candidates

    ///> Start: Constructed affine candidates
    {
      MotionInfo mi[4];
      bool isAvailable[4] = { false };

      int8_t neighGbi[2] = { GBI_DEFAULT, GBI_DEFAULT };
      // control point: LT B2->B3->A2
      const Position posLT[3] = { pu.Y().topLeft().offset( -1, -1 ), pu.Y().topLeft().offset( 0, -1 ), pu.Y().topLeft().offset( -1, 0 ) };
      for ( int i = 0; i < 3; i++ )
      {
        const Position pos = posLT[i];
        const PredictionUnit* puNeigh = cs.getPURestricted( pos, pu, pu.chType );

        if ( puNeigh && CU::isInter( *puNeigh->cu )
          )
        {
          isAvailable[0] = true;
          mi[0] = puNeigh->getMotionInfo( pos );
          neighGbi[0] = puNeigh->cu->GBiIdx;
          break;
        }
      }

      // control point: RT B1->B0
      const Position posRT[2] = { pu.Y().topRight().offset( 0, -1 ), pu.Y().topRight().offset( 1, -1 ) };
      for ( int i = 0; i < 2; i++ )
      {
        const Position pos = posRT[i];
        const PredictionUnit* puNeigh = cs.getPURestricted( pos, pu, pu.chType );


        if ( puNeigh && CU::isInter( *puNeigh->cu )
          )
        {
          isAvailable[1] = true;
          mi[1] = puNeigh->getMotionInfo( pos );
          neighGbi[1] = puNeigh->cu->GBiIdx;
          break;
        }
      }

      // control point: LB A1->A0
      const Position posLB[2] = { pu.Y().bottomLeft().offset( -1, 0 ), pu.Y().bottomLeft().offset( -1, 1 ) };
      for ( int i = 0; i < 2; i++ )
      {
        const Position pos = posLB[i];
        const PredictionUnit* puNeigh = cs.getPURestricted( pos, pu, pu.chType );


        if ( puNeigh && CU::isInter( *puNeigh->cu )
          )
        {
          isAvailable[2] = true;
          mi[2] = puNeigh->getMotionInfo( pos );
          break;
        }
      }

      // control point: RB
#if JVET_P1006_PICTURE_HEADER
      if ( slice.getPicHeader()->getEnableTMVPFlag() )
#else
      if ( slice.getEnableTMVPFlag() )
#endif
      {
        //>> MTK colocated-RightBottom
        // offset the pos to be sure to "point" to the same position the uiAbsPartIdx would've pointed to
        Position posRB = pu.Y().bottomRight().offset( -3, -3 );

        const PreCalcValues& pcv = *cs.pcv;
        Position posC0;
        bool C0Avail = false;

        if ( ((posRB.x + pcv.minCUWidth) < pcv.lumaWidth) && ((posRB.y + pcv.minCUHeight) < pcv.lumaHeight) )
        {
          int posYInCtu = posRB.y & pcv.maxCUHeightMask;
          if (posYInCtu + 4 < pcv.maxCUHeight)
          {
            posC0 = posRB.offset(4, 4);
            C0Avail = true;
          }
        }

        Mv        cColMv;
        int       refIdx = 0;
        bool      bExistMV = C0Avail && getColocatedMVP( pu, REF_PIC_LIST_0, posC0, cColMv, refIdx, false );
        if ( bExistMV )
        {
          mi[3].mv[0] = cColMv;
          mi[3].refIdx[0] = refIdx;
          mi[3].interDir = 1;
          isAvailable[3] = true;
        }

        if ( slice.isInterB() )
        {
          bExistMV = C0Avail && getColocatedMVP( pu, REF_PIC_LIST_1, posC0, cColMv, refIdx, false );
          if ( bExistMV )
          {
            mi[3].mv[1] = cColMv;
            mi[3].refIdx[1] = refIdx;
            mi[3].interDir |= 2;
            isAvailable[3] = true;
          }
        }
      }

      //-------------------  insert model  -------------------//
      int order[6] = { 0, 1, 2, 3, 4, 5 };
      int modelNum = 6;
      int model[6][4] = {
        { 0, 1, 2 },          // 0:  LT, RT, LB
        { 0, 1, 3 },          // 1:  LT, RT, RB
        { 0, 2, 3 },          // 2:  LT, LB, RB
        { 1, 2, 3 },          // 3:  RT, LB, RB
        { 0, 1 },             // 4:  LT, RT
        { 0, 2 },             // 5:  LT, LB
      };

      int verNum[6] = { 3, 3, 3, 3, 2, 2 };
      int startIdx = pu.cs->sps->getUseAffineType() ? 0 : 4;
      for ( int idx = startIdx; idx < modelNum; idx++ )
      {
        int modelIdx = order[idx];
        getAffineControlPointCand(pu, mi, isAvailable, model[modelIdx], ((modelIdx == 3) ? neighGbi[1] : neighGbi[0]), modelIdx, verNum[modelIdx], affMrgCtx);
        if ( affMrgCtx.numValidMergeCand != 0 && affMrgCtx.numValidMergeCand - 1 == mrgCandIdx )
        {
          return;
        }

        // early termination
        if ( affMrgCtx.numValidMergeCand == maxNumAffineMergeCand )
        {
          return;
        }
      }
    }
    ///> End: Constructed affine candidates
  }

  ///> zero padding
  int cnt = affMrgCtx.numValidMergeCand;
  while ( cnt < maxNumAffineMergeCand )
  {
    for ( int mvNum = 0; mvNum < 3; mvNum++ )
    {
      affMrgCtx.mvFieldNeighbours[(cnt << 1) + 0][mvNum].setMvField( Mv( 0, 0 ), 0 );
    }
    affMrgCtx.interDirNeighbours[cnt] = 1;

    if ( slice.isInterB() )
    {
      for ( int mvNum = 0; mvNum < 3; mvNum++ )
      {
        affMrgCtx.mvFieldNeighbours[(cnt << 1) + 1][mvNum].setMvField( Mv( 0, 0 ), 0 );
      }
      affMrgCtx.interDirNeighbours[cnt] = 3;
    }
    affMrgCtx.affineType[cnt] = AFFINEMODEL_4PARAM;

    if ( cnt == mrgCandIdx )
    {
      return;
    }
    cnt++;
    affMrgCtx.numValidMergeCand++;
  }
}

void PU::setAllAffineMvField( PredictionUnit &pu, MvField *mvField, RefPicList eRefList )
{
  // Set Mv
  Mv mv[3];
  for ( int i = 0; i < 3; i++ )
  {
    mv[i] = mvField[i].mv;
  }
  setAllAffineMv( pu, mv[0], mv[1], mv[2], eRefList );

  // Set RefIdx
  CHECK( mvField[0].refIdx != mvField[1].refIdx || mvField[0].refIdx != mvField[2].refIdx, "Affine mv corners don't have the same refIdx." );
  pu.refIdx[eRefList] = mvField[0].refIdx;
}

void PU::setAllAffineMv(PredictionUnit& pu, Mv affLT, Mv affRT, Mv affLB, RefPicList eRefList, bool clipCPMVs)
{
  int width  = pu.Y().width;
  int shift = MAX_CU_DEPTH;
  if (clipCPMVs)
  {
    affLT.mvCliptoStorageBitDepth();
    affRT.mvCliptoStorageBitDepth();
    if (pu.cu->affineType == AFFINEMODEL_6PARAM)
    {
      affLB.mvCliptoStorageBitDepth();
    }
  }
  int deltaMvHorX, deltaMvHorY, deltaMvVerX, deltaMvVerY;
  deltaMvHorX = (affRT - affLT).getHor() << (shift - floorLog2(width));
  deltaMvHorY = (affRT - affLT).getVer() << (shift - floorLog2(width));
  int height = pu.Y().height;
  if ( pu.cu->affineType == AFFINEMODEL_6PARAM )
  {
    deltaMvVerX = (affLB - affLT).getHor() << (shift - floorLog2(height));
    deltaMvVerY = (affLB - affLT).getVer() << (shift - floorLog2(height));
  }
  else
  {
    deltaMvVerX = -deltaMvHorY;
    deltaMvVerY = deltaMvHorX;
  }

  int mvScaleHor = affLT.getHor() << shift;
  int mvScaleVer = affLT.getVer() << shift;

  int blockWidth = AFFINE_MIN_BLOCK_SIZE;
  int blockHeight = AFFINE_MIN_BLOCK_SIZE;
  const int halfBW = blockWidth >> 1;
  const int halfBH = blockHeight >> 1;

  MotionBuf mb = pu.getMotionBuf();
  int mvScaleTmpHor, mvScaleTmpVer;
  const bool subblkMVSpreadOverLimit = InterPrediction::isSubblockVectorSpreadOverLimit( deltaMvHorX, deltaMvHorY, deltaMvVerX, deltaMvVerY, pu.interDir );
  for ( int h = 0; h < pu.Y().height; h += blockHeight )
  {
    for ( int w = 0; w < pu.Y().width; w += blockWidth )
    {
      if ( !subblkMVSpreadOverLimit )
      {
        mvScaleTmpHor = mvScaleHor + deltaMvHorX * (halfBW + w) + deltaMvVerX * (halfBH + h);
        mvScaleTmpVer = mvScaleVer + deltaMvHorY * (halfBW + w) + deltaMvVerY * (halfBH + h);

      }
      else
      {
        mvScaleTmpHor = mvScaleHor + deltaMvHorX * ( pu.Y().width >> 1 ) + deltaMvVerX * ( pu.Y().height >> 1 );
        mvScaleTmpVer = mvScaleVer + deltaMvHorY * ( pu.Y().width >> 1 ) + deltaMvVerY * ( pu.Y().height >> 1 );
      }
      roundAffineMv( mvScaleTmpHor, mvScaleTmpVer, shift );
      Mv curMv(mvScaleTmpHor, mvScaleTmpVer);
      curMv.clipToStorageBitDepth();

      for ( int y = (h >> MIN_CU_LOG2); y < ((h + blockHeight) >> MIN_CU_LOG2); y++ )
      {
        for ( int x = (w >> MIN_CU_LOG2); x < ((w + blockWidth) >> MIN_CU_LOG2); x++ )
        {
          mb.at(x, y).mv[eRefList] = curMv;
        }
      }
    }
  }

  pu.mvAffi[eRefList][0] = affLT;
  pu.mvAffi[eRefList][1] = affRT;
  pu.mvAffi[eRefList][2] = affLB;
}

void clipColPos(int& posX, int& posY, const PredictionUnit& pu)
{
  Position puPos = pu.lumaPos();
  int log2CtuSize = floorLog2(pu.cs->sps->getCTUSize());
  int ctuX = ((puPos.x >> log2CtuSize) << log2CtuSize);
  int ctuY = ((puPos.y >> log2CtuSize) << log2CtuSize);
  int horMax = std::min( (int)pu.cs->pps->getPicWidthInLumaSamples() - 1, ctuX + (int)pu.cs->sps->getCTUSize() + 3 );
  int horMin = std::max((int)0, ctuX);
  int verMax = std::min( (int)pu.cs->pps->getPicHeightInLumaSamples() - 1, ctuY + (int)pu.cs->sps->getCTUSize() - 1 );
  int verMin = std::max((int)0, ctuY);

  posX = std::min(horMax, std::max(horMin, posX));
  posY = std::min(verMax, std::max(verMin, posY));
}

bool PU::getInterMergeSubPuMvpCand(const PredictionUnit &pu, MergeCtx& mrgCtx, bool& LICFlag, const int count
  , int mmvdList
)
{
  const Slice   &slice = *pu.cs->slice;
  const unsigned scale = 4 * std::max<int>(1, 4 * AMVP_DECIMATION_FACTOR / 4);
  const unsigned mask = ~(scale - 1);

  const Picture *pColPic = slice.getRefPic(RefPicList(slice.isInterB() ? 1 - slice.getColFromL0Flag() : 0), slice.getColRefIdx());
  Mv cTMv;

  if ( count )
  {
    if ( (mrgCtx.interDirNeighbours[0] & (1 << REF_PIC_LIST_0)) && slice.getRefPic( REF_PIC_LIST_0, mrgCtx.mvFieldNeighbours[REF_PIC_LIST_0].refIdx ) == pColPic )
    {
      cTMv = mrgCtx.mvFieldNeighbours[REF_PIC_LIST_0].mv;
    }
    else if ( slice.isInterB() && (mrgCtx.interDirNeighbours[0] & (1 << REF_PIC_LIST_1)) && slice.getRefPic( REF_PIC_LIST_1, mrgCtx.mvFieldNeighbours[REF_PIC_LIST_1].refIdx ) == pColPic )
    {
      cTMv = mrgCtx.mvFieldNeighbours[REF_PIC_LIST_1].mv;
    }
  }

  ///////////////////////////////////////////////////////////////////////
  ////////          GET Initial Temporal Vector                  ////////
  ///////////////////////////////////////////////////////////////////////
#if !JVET_P0385_UNIFIED_MV_ROUNDING
  int mvPrec = MV_FRACTIONAL_BITS_INTERNAL;
#endif

  Mv cTempVector = cTMv;
  bool  tempLICFlag = false;

  // compute the location of the current PU
  Position puPos = pu.lumaPos();
  Size puSize = pu.lumaSize();
  int numPartLine = std::max(puSize.width >> ATMVP_SUB_BLOCK_SIZE, 1u);
  int numPartCol = std::max(puSize.height >> ATMVP_SUB_BLOCK_SIZE, 1u);