to improve and complete the draft text for MV-HEVC conformance test and 3D-HEVC conformance test
to collect the missing conformance bitstreams for 3D-HEVC
to collect the missing conformance bitstreams for MV-HEVC
to discuss procedures on how to cross verify the 3D bitstreams
Further discussion & BoG (JCT3V-J0111) was held during the week.
Project development, status, and guidance (1)
Communication by parent bodies (0)
Joint session with VCEG & MPEG parent-level and JCT-VC and JCT-3V, chaired by JRO, GJS, & JO Monday 10-20 1600-1800:
Non-4:4:4 for SCC? (e.g., JCTVC-S0042)
The contribution does not request a 4:2:0 profile, but requests the capability to be present in the 4:4:4-capable profile so that encoders can choose to use that.
Some participants indicated that if we think 4:2:0 is important, we should go ahead and define a profile for it.
Agreed: It was suggested to consider 4:4:4 the primary requirement, but try to also consider the potential application of the same coding tools for 4:2:0.
Decision: Adding new 4:2:0 coding capabilities would be adding some complexity to 4:4:4 decoders, but could be considered if the added complexity is not excessive and the benefit is significant – consider the complexity/benefit tradeoff. Profiling implications TBD.
JCT3V-J0106 Profiles, tiers, levels for 3D-HEVC – proposing one profile, very similar to MV-HEVC, with nested MV-HEVC decoding capability. Decision: Adopt (possibly with refinement of low-level aspects), with nested capability.
SEI & VUI & CICP
JCTVC-S0148 End-of-picture indication for AVC & HEVC – OK.
JCTVC-S0031 / JCT3V-J0108 Frame packing for video with depth (and relationship with depth and texture view packing SEI message in AVC) – to be considered in JCT-3V – only interest if there's a significant benefit. JCT3V-J0109 is new proposed draft text, to be reviewed in JCT-3V to determine adequacy of editorial quality for proceeding with the ongoing work together with MFC+D.
JCTVC-S0197 VUI codepoint for SMPTE ST 2085 (YDzDx) for AVC & HEVC – proceed.
ST 2084 and 2086 for AVC – proceed.
m35153 P3 colour primaries in CICP ST 428-2 or maybe -1 – OK to support SMPTE specs.
[Note also JEDEC P22 – TBD]
Green metadata – SEI message referring to another standard for the details.
Unlimited level / unconstrained profile for AVC & HEVC – still plan to do, but no draft yet
Future video coding exploration being conducted (WCG/HDR, compression improvement (workshop), other)
About the texture/depth view packing, JCT-3V shall decide whether the text is mature.
There should not be different texture/depth view packing messages – what are the market needs?
It was also discussed to potentially de-couple the two amendment activities on MFC+D and TDVP SEI.
3D video application areas (0)
No contributions noted in this area.
MV-HEVC / 3D-HEVC Conformance (0)
Joint meeting with JCT-VC
Present the current status of conformance MV-HEVC and SHVC
Talk about possibility of joint development of MV-HEVC & SHVC reference decoder
Some principles thought about in context of SHVC:
Extension to current conformance for new profiles,
Naming of bitstreams for dependent layers
One md5 sum for each layer in the bitstream
For each picture in output layer sets, log file including NUH Layer ID, POC, MD5 checksum
Output capability of this log file from decoder software (i.e. they could be generated automatically from the streams)
List of bitstreams (in many cases volunteers to be identified), current design more for testing tools functionality rather than extreme cases.
MV-HEVC:
Not testing all HL syntax (assuming that SHVC is doing that).
MD5 separate for each output view (not per picture), and for the entire bitstream.
Set of 8 bitstreams defined.
No volunteer identified for hybrid scalability.
Are the bitstreams sufficiently testing the HL syntax?
Additional bitstream testing max number of output views may be necessary.
Naming convention for layers should be aligned with SHVC.
Currently missing in MV-HEVC:software:
Hybrid scalability
Currently missing in SHVC software:
Aux pictures
Encoders would be different – have specific optimization approaches
Currently, no advantage seen in merging the decoder software, but it would be a useful functionality if SHM was capable to decode multiview streams.
Conclusion:
Produce amendments on software and conformance for MV and SHVC independently.
13.0.1.1.1.1.1.1.12JCT3V-J0111 BoG on HEVC Conformance for 3D extensions [Y. Chen]
Review for 3D-HEVC was performed Thursday PM. It was concluded to issue only one conformance draft, both for MV-HEVC and 3D-HEVC. Further discussion of the work plan was performed during the Friday PM plenary – see notes in section 9.3.