Organisation internationale de normalisation

Project development

Joint meetings are discussed in this section of this report.

A joint meeting with the parent bodies (VCEG and MPEG Requirements and Video) was held on Monday 22 June during 16:30-17:00 on SCC, then 17:00-18:00 on HDR

• 
  SCC profiles/levels and limits (coding tools, IBC referencing range, chroma format, bit depth, …)
  Decision:
  
  • 
    Palette size <= 64, palette predictor size <= 128
    
  • 
    Intra block copy referencing range – not restricted
    
  • 
    Suggestion to specify a 4:4:4 8 bit profile and a (nested) 4:2:0 8 bit profile.
    
  • 
    Four profiles: 8 & 10 bit × 4:4:4 & 4:2:0
    
    • 
      The 4:4:4 profile will support 4:2:0 video (in a nested fashion)
      
    • 
      4:2:2 support not to be required
      
  • 
    RExt coding tools supported
    
  • 
    INBLD supported
    
  • 
    Friday closing plenary 12:30: Aspects not mentioned, such as bit rate and CPB size are to be set as in previous comparable profiles – editors have some discretion if aspects are not noted. A participant commented that for 4:4:4 10, they would like support of the CABAC alignment of the high throughput – this aspect not included for now; for further study.
    
• 
  It was noted that MPEG will ballot a DIS, rather than a DAM, for the next stage of balloting for SCC, and the editors were asked to prepare the necessary output text accordingly.
  
• 
  The proposed palette with pixel copy prediction mode (U0116 and U0171) was discussed – no action.
  
• 
  Proposed SEI messages were discussed
  
  • 
    Some were noted to be HDR related
    
  • 
    Ambient viewing environment SEI message (U0112)
    
    • 
      Suggestion: Study ST 2094 and MPEG-V (Marius)
      
  • 
    Video bitstream transition SEI message (U0115) – to be further discussed (joint meeting of JCT-VC and MPEG systems and video on Tuesday 17:30-18:00)
    
• 
  HDR video coding: Upon evaluation of the responses to the MPEG Call for Evidence (CfE) on HDR/WCG video, it was concluded that a substantial commonality existed between the proposed technical approaches and that the rapid standardization of an approach for improved HDR video coding might be feasible to develop without making any changes to the central HEVC decoding process and low-level syntax. Due to the desire to act quickly and the limited scope anticipated for that work, no plan to follow the CfE with a formal Call for Proposals seemed necessary. General support was expressed for allocating to the JCT-VC the technical work to develop the short-term specification, and management-level discussions were to take place toward that action. Additional longer-term work on HDR video coding may also occur in another phase - e.g., as part of a next-generation video coding standardization project.
  

Tuesday 23 June 17:30-18:00 Joint meeting with MPEG Systems and Video

• 
  WG 11 input M36167 from M. M. Hannuksela et al. on dynamic resolution conversion (DRC/RPR) for video streaming had been reviewed by MPEG Systems prior to the joint meeting, and was shortly summarized again (this is a follow-up of docs M35766 and M36037 of the 111th WG 11 meeting) was discussed, no specific action was planned, other than encouraging further study for future video coding developments. Currently with HEVC, switching between streams (and potentially at the same time changing spatial resolution) is possible at IDR/IRAP pictures, which is hardly used in practice (due to overhead with closed GOP IDR pictures). A desire was expressed to allow switching at CRA positions as well (not just at IDR pictures) seamlessly, which is currenty supported only when no resolution change is done. Here it is requested to do this with reference picture resampling (to be specified normatively). The proponents of such technology were asked whether drift occurs, but this was said to be subjectively unnoticeable.
  
• 
  It was also mentioned that such an approach could be implemented by using SHVC (even without drift), where, however, there is some bit rate overhead and the necessity to implement two loops would be undesirable. Some provision for this usage was discussed previously for enabling an indication in high-level syntax that only one resolution layer will need to be decoded in each picture interval (to avoid the extra complexity of two-loop decoding), and this may already be supported in high-level syntax. The proposed solution could also be interpreted as lightweight SHVC, and even could be using the same filter design as for HEVC upsampling, however these would need to be run only at certain picture positions, similar to switching pictures in AVC. Also, although SHVC enables switching up in picture resolution in a predictive manner, it does not enable switching down in resolution predictively.
  
• 
  Conformant decoding would require defining a new profile to support this. Some concern was raised whether this would be justified for this specific case, where it would be questionable if that would ever be widely used. Even without a profile, it was suggested that decoders could be implemented running such a method in a non-conforming way, although this would presumably violate the current standard and cause interoperability problems with decoders not designed for it.
  
• 
  Further study of this is necessary. Several experts mentioned that such a functionality might be desirable in next-generation video coding.
  

Thursday 25 June 14:30-15:15 Joint meeting with VCEG and MPEG Requirements and Video

• 
  For new work on HDR video coding, it was reported that further management-level discussions at the parent-body level needed to take place before the JCT-VC would be given the mandate to work on this topic. Thus, the work toward standardization action on HDR (except as otherwise noted) was to remain at the parent-body level for the time being.
  
• 
  Scalable lossless coding was discussed. A contribution m36373 had been submitted to MPEG by Fujitsu that reported a potential gain of 1–6% over JPEG 2000 for a scalable lossless coding scheme based on HEVC all-intra coding with a lossless enhancement. Further study of this was encouraged.
  
• 
  The ARIB EOTF/OETF U0032 and U0033 were approved to be considered by JCT-VC – see notes on those contributions.
  
• 
  On 4:2:0 chroma positioning for ITU-R BT.2020 video, a planned liaison communication with ITU-R WP6C was discussed (see notes for U0156)
  

3. BoGs

1.1.1.1.1.1.1.1.157JCTVC-U0175 BoG report on worst case memory bandwidth assessment [K. Rapaka]

See section 4.2.1.

1.1.1.1.1.1.1.1.158JCTVC-U0178 BoG report on constrained intra prediction for IBC unification [X. Xiu]

See section 5.1.3.

4.