Commit cf76914b authored by Karsten Suehring's avatar Karsten Suehring

update version to 5.0rc1

parent f6d5d444
Pipeline #1480 passed with stage
#if ! defined( VTM_VERSION )
#define VTM_VERSION "4.2"
#define VTM_VERSION "5.0rc1"
#endif
  • Hi team VVC, Does JVET_N0193_LFNST work with RExt__HIGH_BIT_DEPTH_SUPPORT?

    I have problem with TrQuant.cpp.

    *out++ = Clip3<int>( outputMinimum, outputMaximum, ( int ) ( resi + 64 ) >> 7 );

    invLfnstNxN( (int*) m_tempInMatrix, (int*) m_tempOutMatrix, g_lfnstLut[ intraMode ], lfnstIdx - 1, sbSize, ( tu4x4Flag || tu8x8Flag ) ? 8 : 16 );

    fwdLfnstNxN( (int*) m_tempInMatrix, (int*) m_tempOutMatrix, g_lfnstLut[ intraMode ], lfnstIdx - 1, sbSize, ( tu4x4Flag || tu8x8Flag ) ? 8 : 16 );

    Does EncSlice.cpp work with ENABLE_WPP_PARALLELISM

    I have problem with create object file.

    for( int ctuTsAddr = startCtuTsAddr; ctuTsAddr < boundingCtuTsAddr; ctuTsAddr += pcPic->cs->pcv->widthInCtus )

    encodeCtus( pcPic, bCompressEntireSlice, bFastDeltaQP, ctuTsAddr, ctuTsAddr + pcPic->cs->pcv->widthInCtus, m_pcLib );

    CHECK(cs.sps == 0, "No SPS present");

    writeBlockStatisticsHeader(cs.sps);

    Edited by Jamaika
  • Those macros are not regularly tested. ENABLE_WPP_PARALLELISM was fixed in VTM 4.1. RExt__HIGH_BIT_DEPTH_SUPPORT my generally not be supported.

    Generally bugs should be reported in the bug tracker. There is no index for comments like this, so these are likely to get lost.

  • Thanks for the answer. I'm just an amateur. I satisfied my curiosity. Sometimes it compiles the codec on doom9 forum.

    One more question.

    Is MAIN10 change to MAINREXT in EncAppCfg.cpp for 12bit or chroma subsampling 422/444 is a big mistake? Is the RExt__HIGH_BIT_DEPTH_SUPPORT function required for this change?

    m_bitDepthConstraint = ( ( m_profile == Profile::MAINREXT || m_profile == Profile::NEXT ) ? 12 : 8 );

    Short test additional function for yuv420p 8bit QP=32 CTU=32 LCTUFast=0 with encoder_randomaccess_vtm.cfg

    no function = effect blur

    add full paralleism = much smaller effect blur only frame I

    add tracing = test encoder, longer total time

    add rext = much smaller effect blur only frame I

    add rext + paralleism = effect banding

    add only split paralleism = file is the same as no function

    Conclusions:

    LCTUFast should be enabled for small CTUs. The frames have more details.

    Edited by Jamaika
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment