RCE1 related (high bit rate coding and high bit depth) (2)
14.1.97.1.1.1.1.1.172JCTVC-P0075 non-[RCE1, RCE2]: Combination of RCE1 subtests B5 and B6 with RCE2 subtest A1 [R. Joshi, L. Guo, J. Sole, M. Karczewicz (Qualcomm)]
Presentation not requested.
14.1.97.1.1.1.1.1.173JCTVC-P0289 Non-RCE1: Combination of subtests A1 and B3.a of RCE1 [R. Joshi, J. Sole, M. Karczewicz (Qualcomm)] [late]
Presentation not requested.
Implementation aspects of high bit rate and high bit depth (6)
14.1.97.1.1.1.1.1.174JCTVC-P0044 On MinCR [T. Suzuki (Sony)]
See BoG report P0288 and related notes. Discussed further in plenary 01-16 (GJS).
The constraint on MinCR is specified to limit the worst case of CABAC decoding process. In the current level/tier definition, the value of MinCR is the same for both Main Tier and High Tier. The current MinCR was defined mainly for Main Tier and larger value is used for higher level. For example, MinCR is equal to 8 for level 5.1 (4K@60p). However this value is reportedly too large for high-end professional equipment, and it was suggested to define this as 4. In addition, the implicit minimum sequence compression ratio defined by the maximum bit rate and maximum luma sample rate is asserted to be overly restrictive when using all-intra coding or higher bit depths and chroma formats, to the point where MinCr becomes superfluous. This contribution proposes to reduce the value of MinCR and introduce a new tier to allow increased bit rates for high end professional applications.
Regarding version 1, it was remarked that making such a change to the Main and Main 10 profiles would be a serious matter.
For RExt, it was noted that bit rate is not specified to change with bit depth, but is affected by chroma format.
It was commented that the lack of effect on CPB size in the RExt draft could be justifiable in terms of expecting shorter-GOP and more constant bit rate per picture for the newer profiles.
P0086 is related to this discussion.
Suggestions:
-
P0044 (v6, option 2) has a suggestion with
-
three tiers of 4:4:4 16 b profiles (all-intra 16 b as output of this meeting)
-
scale bit rates proportional to bit depth for new profiles
-
Scale CPB size with bit depth (as well as chroma format)
Decision: Agreed as above (with syntax flag constraining downwards).
14.1.97.1.1.1.1.1.175JCTVC-P0061 AHG18: Worst-case Escape Code Length Mitigation [K. Sharman, N. Saunders, J. Gamei (Sony)]
See BoG report P0288 and related notes.
14.1.97.1.1.1.1.1.176JCTVC-P0296 Cross-check report of 'AHG18: Worst-case Escape Code Length Mitigation' (JCTVC-P0061) by Sony [C. Rosewarne, M. Maeda (Canon)] [late]
14.1.97.1.1.1.1.1.177JCTVC-P0066 AHG5: Luma-chroma prediction for different bitdepth [O. Nakagami, T. Suzuki (Sony)]
Relates to cross-component prediction. See BoG report P0288 and related notes.
14.1.97.1.1.1.1.1.178JCTVC-P0100 AHG5: cross-check results of luma-chroma prediction for different bitdepth (JCTVC-P0066) [K. Kawamura, S. Naito (KDDI)] [late]
14.1.97.1.1.1.1.1.179JCTVC-P0173 AHG 5 and 18: Coding of high bit-depth source with lower bit-depth encoders and a continuity mapping [C. Auyeung, J. Xu (Sony)]
Related to P0162. Was first discussed in RExt BoG P0288, then in HLS BoG., then discussed in plenary 01-16 (GJS).
This contribution proposes a method to encode high bit-depth video by lower bit-depth encoders and a continuity transform. In this contribution, the high bit-depth source is divided into an MSB part and an LSB part. The MSB part is losslessly coded as a primary picture, and the LSB part is transformed by a continuity mapping and then encoded as an auxiliary picture. SEI message is proposed to combine the MSB and the LSB parts, based on the SEI message from JCTVC-O0090. The BDR spreadsheet from the common test conditions, JCTC-L11100, reported that the continuity mapping improved the coding efficiency of the 4 LSB bits of the 10 bit BT.709 test sequences from SCE1 with luma BDR equals to−52.9%, −94.7%, and −93.1% for all intra main, random access main, and low delay B main respectively.
A question asked was what is the relative coding efficiency of this relative to a full-bit-depth encoder (for lossy coding).
Coding results were provided to show that the continuity mapping was beneficial.
It was suggested to consider usage with as many bits as possible placed in the LSBs to test lossy coding efficiency.
Using overlapping ranges and coding the LSBs as a difference from a prediction was suggested as an alternative.
SEI message usage was proposed.
P0162 and O0090 were related. These contributions are interesting for purposes of analysis of coding effectiveness.
This was agreed not to be an action item for RExt.
No action was taken on this.
14.1.97.1.1.1.1.1.180JCTVC-P0287 AhG 5 and 18: Cross-check of coding of high bit-depth source with lower bit-depth encoders and a continuity mapping in JCTVC-P0173 [W.-S. Kim, W. Pu (Qualcomm)] [late]
14.1.97.1.1.1.1.1.181JCTVC-P0222 AhG 18: On SAO performance at high bit-depth and high bit-rates [E. Alshina, A. Tourapis, W.-S. Kim, Y.-W.Huang] [late]
See BoG report P0288 and related notes.
Dostları ilə paylaş: |