(on behalf of Pavel Evsikov pavel.evsikov@…)

In parsePPS, there is

if ((tileColumnsMinus1 + tileRowsMinus1) != 0) {

READ_FLAG ( uiCode, "loop_filter_across_tiles_enabled_flag" ); pcPPS->setLoopFilterAcrossTilesEnabledFlag( uiCode ? true : false );

}

Though (tileColumnsMinus1 + tileRowsMinus1) != 0 is a conformance condition, the spec does not tell that reading of the flag is possible only when both reading parameters are not zeros. The

It is better to user "assert" for the conformance condition.

Bitstreams produced by the hm encoder are not conformant.

In seq_parameter_set_rbsp, and pic_parameter_set_rbsp when the extension is being parsed. The spec says

sps_range_extension_flag u(1) sps_multilayer_extension_flag u(1) sps_3d_extension_flag u(1) sps_scc_extension_flag u(1) sps_extension_4bits u(4)

However in the hm code looking at enum SPSExtensionFlagIndex in TLibCommon/TypeDef.h the reference code parses sps_scc_extension at index 6 in this octet, and not at index 3. This influences decoding/encoding through TDecCavlc::parseSPS/TEncCavlv::codeSPS. Similarly for the PPS.

In 7.3.8.10 Transform unit syntax. ... tu_residual_act_flag ae(v) ...

In 7.4.9.10 Transform unit semantics. tu_residual_act_flag[x0][y0] ...

Syntax table is missing array subscripts.

In palette syntax and semantics, travScan[·][·] is used, this should be ScanOrder[log2BlockSize][3][·][·], also ScanOrder in the semantics should be updated to include travScan.

In 8.5.2.3.1 General, repeated condition

• nPbSw is equal to 8

Also

• noIntegerMV is equal to 1

should be

• noIntegerMv is equal to 1

In equation (8-108) the variables xCurr, yCurr are used, but aren't defined, it probably means xCb, yCb.

In 8.4.2 Derivation process for luma intra prediction mode, candModeList[mpm_idx], should be candModeList[mpm_idx[xPb][yPb]].