International organisation for standardisation organisation internationale de normalisation

Other topics for version 1

See JCTVC-L0363 regarding:

• 
  Contouring artefact
  
• 
  CABAC initialization
  
• 
  POC versus reference index usage.
  

5. Range extensions

   1. General

4.2.0.1.1.1.1.1.123JCTVC-L0445 BoG report on range extensions [D. Flynn]

Break out group discussions were held on the 15th, 16th, 17th and 22nd of January to examine contributions related to range extensions and discuss modifications to the common test conditions.

Contributions covered topics on transform design, intra prediction, colour transforms, in-loop filtering, mixed chroma formats, lossless coding tools and other informative matters. Contributions that proposed modification to low-level syntax were recommended not to be adopted on the basis of small coding benefits at the cost of diverging from the base system. A number of contributions, both in categories of lossless and non-lossless coding, presented measurable coding benefits in the areas of intra prediction and colour space conversion. It was recommended to consider forming CEs to investigate these benefits.

The results of the CE1 investigation on rectangular versus square transforms and additional results from non-CE1 contributions were presented. After eliminating some variants, further discussion was inconclusive without answering the wider question of what the objectives of the range extension are. A discussion on this matter was held and was recommended to be reviewed in the wider group.

From discussion in plenary on lossless coding: It was unclear where this will fit. The group agreed to seek clarification with parent bodies on major application domains of such lossless tools.

Decision: Draft (ISO/IEC PDAM) on Range Extensions (changes relative to the previous WD):

• 
  Square Transform and square intra prediction blocks
  
• 
  Intra prediction angle adjustment for 4:2:2
  
• 
  Software is to include both square and non-square transform schemes
  

There was some discussion about the possibility to include both square and non-square variations in the draft, but several experts expressed that this would be undesirable.

Plenary review was held Wednesday at noon (in Popov).

The remaining informative contributions on test material (L0302, L0428) were reviewed.

L0455 was reviewed. The contribution described an alternative decoding process. It was recommended for further study.

There was no presenter available for the informational document L0332, which is available for further study.

Recommendations for common test conditions were reviewed and approved, including some testing of the 40+ dB range.

Previous meetings have tried to keep the work on the range extensions as minor modifications to support alternative sampling formats. However, there is a conflict between examining the new coding tools and adherence to an informal philosophy that was used to guide the selection of initial modifications.

The following three points originate from a discussion that was chaired by P. Topiwala in the absence of D. Flynn:

1. 
   What is the goal of range extensions? Presumably tools that provide substantial gains over AVC High 4:4:4 profile?
   

Higher bit-depth (12b, 14b), High Resolution (4K, 8K).

2. 
   Can we get new tools that offer substantial gains for these formats. Colour spaces and perhaps sampling filters may be worth studying, and interaction between tools we have.
   

Are two tiers desirable – Consumer and Professional?

Comment: To justify two tiers, there should be a substantial difference in coding efficiency.

Ex: Wireless network connectivity application (PC to projector say), at 50 Mb/s. Q. Is this a consumer or Professional?

3. 
   How to balance consistency with v1 and deeper gains? Consumer more in line with v1, Pro distinct?
   

Suggestion (with general informal agreement in BoG): For now, follow this two-tier approach, and see if, after a couple of meetings, whether that difference is justified. Call them Main and High Tiers.

Examples from discussions (DF):

• 
  Transform & (intra) prediction: Square or non-square? In a simple system it may be desirable to keep a square system, but the performance benefit of the non-square may be useful in higher bitrate systems.
  
• 
  Mixed formats: it is suggested to investigate dual mode 4:4:4--4:2:0 systems, which tier should these belong to? In terms of complexity should it be high? But this has a greater potential use in main tier applications.
  

2. Transforms (CE1 related)

4.2.0.1.1.1.1.1.124JCTVC-L0149 Non-CE1: Test 3 - Transform skip for chroma in 4:2:2 Range Extension [C. Rosewarne, V. Kolesnikov, M. Maeda (Canon)]
4.2.0.1.1.1.1.1.125JCTVC-L0409 Crosscheck of non-CE1 test 3 with transform skip for chroma in 4:2:2 Range Extension (JCTVC-L0149) [J. Sole (Qualcomm)] [late]
4.2.0.1.1.1.1.1.126JCTVC-L0150 Non-CE1: Test 3 - Square transform deblocking for 4:2:2 [C. Rosewarne, V. Kolesnikov, M. Maeda (Canon)]
4.2.0.1.1.1.1.1.127JCTVC-L0424 Cross-check for JCTVC-L0150 [Marta Mrak, Andrea Gabriellini (BBC)] [late]
4.2.0.1.1.1.1.1.128JCTVC-L0151 Non-CE1: Test 3 - 32x32 transform in chroma for 4:2:2 [V. Kolesnikov, C. Rosewarne, M. Maeda (Canon)]
JCTVC-L0417 Crosscheck of non-CE1 test 3 with 32x32 transform in chroma for 4:2:2 Range Extension (JCTVC-L0151) [J. Sole (Qualcomm)] [late]
4.2.0.1.1.1.1.1.129JCTVC-L0351 Non-CE1: Square transform blocks for 4:2:2 [J. Sole, R. Joshi, M. Karczewicz (Qualcomm), A. Gabriellini, M. Mrak (BBC)]
4.2.0.1.1.1.1.1.130JCTVC-L0407 Cross-check of Non-CE1: Square transform blocks for 4:2:2 (L0351) by Qualcomm [C. Rosewarne, M. Maeda (Canon)] [late]

3. Intra prediction

4.2.0.1.1.1.1.1.131JCTVC-L0112 AHG7: Reference sample filtering in intra prediction for chroma extended format [J. Min, C. Kim (Samsung)]
4.2.0.1.1.1.1.1.132JCTVC-L0391 Crosscheck report of reference sample filtering in intra prediction for chroma extended format (JCTVC-L0112) [S. Matsuo, S. Takamura, H. Fujii, A. Shimizu (NTT)] [late]
4.2.0.1.1.1.1.1.133JCTVC-L0128 AHG7: Modification of intra angular prediction blending [S. Matsuo, S. Takamura, H. Fujii, A. Shimizu (NTT)]
4.2.0.1.1.1.1.1.134JCTVC-L0374 Cross-check of modified intra angular blending (JCTVC-L0128) [K. Kawamura, S. Naito (KDDI)] [late]
4.2.0.1.1.1.1.1.135JCTVC-L0129 AHG7: Modification filter for intra angular prediction [S. Matsuo, S. Takamura, H. Fujii, A. Shimizu (NTT)]
4.2.0.1.1.1.1.1.136JCTVC-L0375 Cross-check of Modification filter for intra angular prediction (JCTVC-L0129) [K. Kawamura, S. Naito (KDDI)] [late]
4.2.0.1.1.1.1.1.137JCTVC-L0399 AHG7: Cross-check results of Sample-based angular intra prediction for HEVC range extension (JCTVC-L0176) [W. Gao, J. Ye, H. Yu (Huawei)] [late]

4. Mixed format coding

4.2.0.1.1.1.1.1.138JCTVC-L0162 AHG7: Coding 4:2:2 chroma format with 4:2:0 and 4:4:4 format codecs [A. Gabriellini, M. Mrak (BBC)]
4.2.0.1.1.1.1.1.139JCTVC-L0436 AHG7: Cross-check of coding 4:2:2 chroma format with 4:2:0 and 4:4:4 format codecs in JCTVC-L0162 [K. Sharman, N. Saunders, J. Gamei (Sony)] [late]
4.2.0.1.1.1.1.1.140JCTVC-L0175 AHG7: A Consumer 4:4:4 Coding Method Based On A 4:2:0 Codec [P. Topiwala, W. Dai, M. Krishnan (FastVDO)]
4.2.0.1.1.1.1.1.141JCTVC-L0321 AHG7: Cross-verification of JCTVC-L0175, A Consumer 4:4:4 Coding Method Based On A 4:2:0 Codec [R. Cohen (MERL)] [late]
4.2.0.1.1.1.1.1.142JCTVC-L0240 AHG7: The performance of extended chroma mode for non 4:2:0 format [J.Kim (LG)]
4.2.0.1.1.1.1.1.143JCTVC-L0355 Crosscheck of LG's proposal JCTVC-L0240 [Xingyu Zhang (HKUST), Wenjing Zhu, Oscar C. Au] [late]
4.2.0.1.1.1.1.1.144JCTVC-L0250 Mixed chroma format coding [M. Budagavi, M. Zhou (TI)]

5. Other

4.2.0.1.1.1.1.1.145JCTVC-L0127 AHG7: LCU-based non-local means filter for chroma components in HEVC range extensions [M. Matsumura, S. Takamura, H. Fujii, A. Shimizu (NTT)]
4.2.0.1.1.1.1.1.146JCTVC-L0373 Cross-check of non-local means filter for chroma components (JCTVC-L0127) [K. Kawamura, S. Naito (KDDI)] [late]
4.2.0.1.1.1.1.1.147JCTVC-L0370 AHG7: Inter-plane intra coding of residual signals for range extensions [K. Kawamura, T. Yoshino, S. Naito (KDDI)] [late]
4.2.0.1.1.1.1.1.148JCTVC-L0123 Cross-check report of inter-plane intra coding of residual signals (JCTVC-L0370) [S. Matsuo, M. Matsumura, S. Takamura, H. Fujii, A. Shimizu (NTT)] [late]
4.2.0.1.1.1.1.1.149JCTVC-L0371 AHG7: In-loop color-space transformation of residual signals for range extensions [K. Kawamura, T. Yoshino, S. Naito (KDDI)] [late]
4.2.0.1.1.1.1.1.150JCTVC-L0124 Cross-check report of in-loop color-space transformation of residual signals (JCTVC-L0371) [S. Matsuo, M. Matsumura, S. Takamura, H. Fujii, A. Shimizu (NTT)] [late]
4.2.0.1.1.1.1.1.151JCTVC-L0189 Analysis on high internal bit depth processing [R. Weerakkody, A. Gabriellini, M. Naccari, M. Mrak (BBC)]
JCTVC-L0455 Cross-segment decoding of HEVC bitstreams with SAD/PSNR information [Gene Wen, Shunyao Li, Spencer Cheng] [late]

(Information document – relates to system environment with video/systems interaction.)

6.