International organisation for standardisation organisation internationale de normalisation

Reconfigurable Media Coding – Video related

4.2.0.1.1.1.1.1.73m28171 Proposal of a Decoder Energy Management Scheme with RVC [Eduardo Juarez, Rong Ren, Jianguo Wei, Mickael Raulet, Matias Javier Garrido, Cesar Sanz, Fernando Pescador]

This proposal suggests to use information about platform and RVC-based decoder configuration to determine energy consumption. Currently the local sequence characteristics (bit rate, motion etc) are not taken into consideration.

No specific video related action was taken, but a connection with the "Green MPEG" concept is apparent.

4.2.0.1.1.1.1.1.74m28209 RVC description of a low delay HEVC decoder [M. Raulet, G. Cocherel, M. Mattavelli, D. De Saint Jorre, E. Juarez]

Implementation of Intra-only, LD-P, LD-B. Deblocking filters and SAO were tested separately so far. It was suggested to wait for HM10 conformance streams to be developed. The scheme reportedly decodes WVGA sequences with approx. 25 fps (2 cores). Tiles are not implemented yet, WPP is partially implemented. The CABAC engine is the same, but the FU entropy decoder is different. The CU/TU tree was reportedly not easy to implement.

It was decided to issue a WD (N13337) and to start a CE on further improvements (N13338).

8. HEVC

In coordination with JCT-VC and JCT-3V which had met in parallel under ITU-T auspices, the following documents were approved as outputs:

• 
  N13341 Disposition of Comments on ISO/IEC DIS 23008-2
  
• 
  N13331 ISO/IEC FDIS 23008-2 High Efficiency Video Coding
  
• 
  N13345 ISO/IEC 23008-2:201x/PDAM1 HEVC Range Extensions
  
• 
  N13343 High Efficiency Video Coding (HEVC) Test Model 10 (HM 10) Encoder Description
  
• 
  N13347 ISO/IEC 23008-2:201x/PDAM2 HEVC Multiview Extensions
  
• 
  N13348 3D-HEVC Test Model 3
  
• 
  N13349 Scalable HEVC (SHVC) Working Draft 1
  
• 
  N13350 Scalable HEVC (SHVC) Test Model 1 (SHM 1)
  
• 
  N13351 Request for Subdivision: ISO/IEC 23008-5
  
• 
  N13352 ISO/IEC CD 23008-5 HEVC Conformance Testing and Reference Software
  

The requirements-related issues were discussed in a joint meeting with MPEG Requirements, Monday 1500-1700. The outcome of that joint discussion is included below as noted.

1. Main 10 profile

The following NB comments and other contributions were all supportive of adding the specification of a "Main 10" Profile which had been included in the Study of FDIS. No specific further action was necessary.

4.2.0.1.1.1.1.1.75m27581 Main 10 Still Picture Profile [C. Auyeung, A. Tabatabai, T. Suzuki (Sony)]

In the joint discussion with Requirements, it was noted that this contribution proposes to specify a 10 bit still picture profile – specified similarly as the Main Still Picture profile. This was not necessarily suggested to be included in version 1 of HEVC. It was clearly agreed not to do this in version 1.

It was commented that perhaps we should wait until Rext progresses, and perhaps such a profile should also contain other features such as 4:4:4, and perhaps also higher bit depths than 10 bit.

It was further commented that having metadata along with the coded picture would be useful.

It was agreed that we need a more clear requirements study on this topic before proceeding, and that we should collaborate with JPEG to hear if they have input to provide on the subject.

4.2.0.1.1.1.1.1.76m27696 FRNB Positions on HEVC [Pierrick Philippe on behalf of the FRNB]
4.2.0.1.1.1.1.1.77m28030 HEVC Main 10 profile [Per Fröjdh on behalf of SENB]
4.2.0.1.1.1.1.1.78m28095 ITNB comment on Main 10 Profile in HEVC [IT National Body]
4.2.0.1.1.1.1.1.79m28056 UKNB Comment on Main 10 Profile of HEVC [UK National Body] [late]

2. HEVC Version 1 and Range Extensions

4.2.0.1.1.1.1.1.80m28033 CANNB comments on HEVC [late]

This contribution indicates satisfaction with the specification of the Main 10 profile, and requests not to plan to create another 10 bit 4:2:0 profile in the absence of strong justification. No action was needed on this, as our plans are already in accord with the advocated statements.

The contribution also requested a 6-month longer timeline for RExt development (i.e. completion in June 2014 rather than January 2014), in consideration of the inclusion of specification of the Main 10 profile in version 1 which can fulfil some of the short-term needs that were envisioned when the RExt schedule was established. The contribution proposed to issue a PDAM in July 2013 rather than January 2013.

In the joint discussion with Requirements, it was noted that UK has wanted rapid progress, but has not reviewed the suggestion.

Currently Rext focuses on 4:4:4 and 4:2:2 (as WD content).

It was suggested that the compression performance may have issues (esp. beyond 10 b) – i.e. that the gain over AVC is reportedly minimal at very high bit rates / quality and that additional coding tools are under study.

In the joint discussion with Requirements, it was decided not to accomodate such a request at this time. The progress of work can be further assessed at subsequent meetings.

The contribution also requested to move SEI messages to a separate part of the standard. However, the concerns expressed were adequately addressed by the other actions taken in JCT-VC, including in-line specification of SEI messages rather than referencing of AVC and the review of which aspects of SEI messages would be included.

4.2.0.1.1.1.1.1.81m28319 Supporting document for CANNB comments on the Frame packing arrangement SEI message

The procedural concerns expressed in this contribution were properly addressed by actions of JCT-VC.

3. HEVC Lossless Coding

4.2.0.1.1.1.1.1.82m28330 Request for developing a lossless coding profile for graphics and text with motion [H. Yu, R. Cohen, A. Luthra, T. Lin, W. Wan, P. Topiwala, A. Segall]

This contribution was discussed in the Monday joint meeting session with the Requirements SG.

The contribution included an emphasis on 4:4:4 graphics and text with motion – e.g. remote desktop and screen sharing.

A suggestion was to have separate coding of the video part if the content is mixed. It was noted that sometimes the content has been pre-combined. The issue of system-level segmentation versus video-level segmentation was brought up. An alternative is to have a single codec for both text/graphics and camera-view video within the coding layer representation. It was remarked that this could possibly have region-adaptive switching between 4:4:4 and 4:2:0.

The concepts and possibilities of "near lossless" or "ordinarily lossless" or "visually lossless" were also discussed.

A suggestion was that a unified design that includes both types of coding tools could support either mixed content, purely graphics & text content, etc.

The current HEVC levels specification design might need some kind of adjustment – e.g. w.r.t. speed and picture size requirements.

There is a need to flesh out the application requirements to clarify what is needed.

A better understanding about applications and requirements is to be developed until next meeting.

4.2.0.1.1.1.1.1.83m27647 AHG8: Lossless coding test results for the 4:4:4 screen content sequences [H. Yu (Huawei), W. Gao, J. Ye, Y. Cao, X. Wang]

4.