SHVC profiles and levels (4)
14.1.97.1.1.1.1.1.107JCTVC-P0127 AHG14: On a CGS profile for SHVC [P. Andrivon, P. Bordes, E. François (Technicolor), X. Ducloux (Thomson Video Networks), Y. Ye (Interdigital), A. Duenas (NGCodec), A. Segall (Sharp Labs), Kemal Ugur (Nokia), Xiang Li (Qualcomm), Elena Alshina (Samsung)]
This contribution requests to define a profile for the HEVC scalable extension supporting colour gamut and bit-depth scalability (CGS). Such a scalability technology may be considered in applications standardization committees such as DVB CM-AVC/CM-UHDTV or ATSC3.0. Considering the foreseeable evolution of capture and display device capabilities and use cases, it was asserted that CGS appears to be one major feature to be supported in upcoming mainstream applications. This contribution also suggests to conduct liaison communication on colour gamut scalability requirements and performance between DVB CM-AVC/CM-UHDTV, SC 29/WG 11 and SG 16.
See action recorded from joint discussions with the parent level in section 7.1.
Decision: Scalable Main 10 Profile will be as in scalable main and allowing 10 bit in base and ELs.
14.1.97.1.1.1.1.1.108JCTVC-P0134 Strawman SHVC level constraints [J. Boyce (Vidyo)]
Discussed 01-16 (GJS).
The current SHM draft specification does not define level constraints for SHVC. "Strawman" Scalable Main profile-specific level constraints are proposed. All of the existing general level constraints would apply individually to each layer within the operation point corresponding to the output layer set conforming to the profile/tier/level. In addition, cumulative constraints for all decoded layers are imposed for total picture size, total sample rate and total bit rate, equal to 2x their respective level limits for Main profile.
Decision: Adopt.
Further editorial work may be needed for expressing the bit rate limit.
It was noted that it is necessary to also develop a specification of levels for an enhancement-only profile.
14.1.97.1.1.1.1.1.109JCTVC-P0142 On SHVC level limits [M. M. Hannuksela (Nokia)]
Discussed 01-16 (GJS).
Proposal #1: It is proposed to constrain the value of max_vps_dec_pic_buffering_minus1[ i ][ k ][ j ] as follows:
-
max_vps_dec_pic_buffering_minus1[ i ][ k ][ j ] shall be in the range of 0 to MaxDpbSize − 1 (as specified in clause A.4), inclusive.
It is asserted that previously the sub-DPB sizes were not constrained at all.
Proposal #2:
-
#2a: It is proposed to add a profile-specific constraint (for the Scalable Main profile and the potential Scalable AVC profile) on cross-layer DPB usage.
-
#2b: The value of the constraint is proposed to be equivalent of allowing MaxDpbSize pictures in the sub-DPBs of a two-layer bitstream with 1.5x scalability.
It is asserted that this constraint leaves encoders the flexibility to support any number of layers from 2 to 8, inclusive, for the scalable profiles, while provides a smaller upper limit for the DPB memory usage than the current limit, which is reportedly 8 * MaxDpbSize, where MaxDpbSize is asserted to be 6 for typical picture sizes used under the levels. When quality scalability is in use, all layers may have the same picture width and height and hence the current upper limit of the DPB memory usage is asserted to be 48 pictures in typical picture sizes. With the proposed constraint, the upper limit of the DPB memory usage is asserted to be equivalent to (1 + 2/3 * 2/3) * MaxDpbSize, i.e. about 9 pictures in typical picture sizes (in units of the highest enhancement layer pictures).
Illustrative figures were added in version 2.
JCTVC-P0192 is related; see notes on that contribution.
14.1.97.1.1.1.1.1.110JCTVC-P0203 Hybrid codec scalability profile in SHVC [J. Samuelsson, J. Enhorn, R. Sjöberg (Ericsson)]
This contribution proposes to include a hybrid codec scalability profile as described in JCTVC-O1012 into the SHVC draft with the following modifications:
-
To remove the option of encapsulating AVC NAL units in HEVC NAL units (and just keep the two options of no encapsulation and encapsulating HEVC NAL units in AVC NAL units).
-
To specify that the base layer must obey all constraints specified for the High profile in the AVC specification.
The contribution asserts that only one encapsulation format is needed and that it is important that the AVC NAL units are unmodified (i.e. no additional header is put in front of the AVC NAL unit header).
See also sections 6.4.3 and 7.1.
Dostları ilə paylaş: |