3.7Source video test material
Reviewed in closing plenary.
JCTVC-M0191 Medical imaging sequences for HEVC development [P. Amon, A. Hutter, U.-E. Martin (Siemens)]
This contribution proposes new test sequences from the medical domain for the development of HEVC, especially for the specification of range extensions. The proposed set contains 8-bit, 12-bit, and 16-bit monochrome image data as well as 8-bit RGB content.
Sequence availability status had not been finalized. Further contribution expected at next meeting.
It was commented that having 16 b material would be especially appreciated and that availability well in advance of the next meeting would also be highly desirable.
JCTVC-M0431 YUV444 test sequences for screen content [Wenpeng Ding, Yunhui Shi, Baocai Yin (??)] (BJUT)] [late]
This contribution describes and introduces three YUV 4:4:4 and RGB 4:4:4 test sequences for screen content which are corresponding to the sequences in JCTVC-F741.
Three video sequences will be provided for screen content. The table below tabulates the specification of the sequences.
Sequences
|
Resolution
|
Frames
|
Frame rate fps
|
Chrome format
|
Bit-depth
Per
Sample
|
Doc
|
1280x720
|
500
|
10
|
4:4:4
|
8
|
Slide
|
1280x720
|
500
|
20
|
4:4:4
|
8
|
Web
|
1280x720
|
500
|
10
|
4:4:4
|
8
|
Sequences
|
Brief Description
|
Doc
|
A word document is being editing.
|
SlideShow
|
A slide show with all kinds of animations.
|
Web
|
Web browsing activities.
|
It was remarked that SlideShow is a 4:4:4 version from which the same sequence previously provided in 4:2:0 was derived.
The sequences will be made available on the Univ. Hannover ftp site that is accessible to JCT-VC members.
The availability of the additional content was highly appreciated by the JCT-VC.
4Core experiments in Range Extensions 4.1RCE1: Inter-component decorrelation methods 4.1.1RCE1 summary and general discussion
JCTVC-M0026 RCE1: Summary report of HEVC Range Extensions Core Experiment 1 on Inter component decorrelation methods [J. Sole, K. Kawamura, J. Kim]
A summary of RCE1 on Inter component decorrelation methods for HEVC Range Extensions is reported. Four methods and combinations have been evaluated based on the range extensions common test conditions according to the CE description in JCTVC-L1121.
Test 1.1: LM Chroma
-
JCTVC-M0097 (JCTVC-L0240), “RCE1: The performance of extended chroma mode for non 4:2:0 format”, J. Kim (LGE)
-
Cross-check: JCTVC-M0084 (Samsung)
Test 1.2: Inter-plane intra coding of residual
-
JCTVC-M0410 (JCTVC-L0370), “RCE1: Results of in-loop color-space transformation of residual signals”, K. Kawamura, T. Yoshino, S. Naito (KDDI)
-
Cross-check: JCTVC-M0085 (Samsung)
Test 2.1: In-loop color-space transformation of residual
-
JCTVC-M0411 (JCTVC-L0371), “RCE1: Results of in-loop color-space transformation of residual signals”, K. Kawamura, T. Yoshino, S. Naito (KDDI)
-
Cross-check: Results provided by Canon (JCTVC-M0387)
Test 2.2: Fixed colour transforms
Test 3: Combination of proposals
-
JCTVC-M0049, “Combined Chroma Tools For Range Extensions, Test 3.3”, P. Topiwala (FastVDO) , J. Kim (LG), K. Kawamura (KDDI)
-
Cross-check: JCTVC-M0367 (Qualcomm)
Related non-CE contributions
-
JCTVC-M0116, “Non-RCE1: Multiple LM chroma modes”, C.-Y. Chen, C.-W. Hsu, C.-Y. Tsai, Y.-W. Huang, S. Lei (MediaTek)
-
JCTVC-M0230, “Non-RCE1/Non-RCE2/AHG5/AHG8: Adaptive Inter-Plane Prediction for RGB Content”, T. Nguyen, A. Khairat, D. Marpe (Fraunhofer HHI)
Some issues were noted:
-
WD text was not uploaded with most proposals
-
For Test 1.2, cross-check results do not match for few sequences in bitrates with the proponent, but match among cross-checkers.
-
For Tests 1.2 and 2.1, the proposal results were uploaded late (18/04)
Discussion from report:
-
BD-rate gains of tools reducing the colour correlation on different domain (pixel/residual) are not additive
-
LM chroma (test 1.1) and RM (1.2) gains are not additive (3.1)
-
Largest BD-rate gains for RGB sequences are obtained with fixed colour transforms
-
One non-CE1 contribution in this meeting proposed variants on LM chroma (JCTVC-M0116) and one on the inter-plane prediction for RGB content (JCTVC-M0230).
Test 1.1: LM Chroma (JCTVC-M0097)
-
Two methods for chroma prediction which predicts chroma samples using linear combination of luma samples (LM mode)
-
LM as was proposed in HEVC version 1, extended to non-4:2:0
-
LM where parameter calculation is performed on a CU basis rather than per-TU
Test 1.2: Inter-plane intra coding of residual (JCTVC-M0410, uploaded on 4/18)
-
Inter-plane intra coding of residual signals (RM). In RM, correlation of residual signals between planes is reduced by using the linear model. The prediction parameter of the model is estimated at the encoder, differentially coded by predicting it from neighbouring parameters, and then signalled as additional information.
Test 2.1: In-loop color-space transformation of residual (JCTVC-M0411, uploaded on 4/18)
-
In-loop color-space transformation of residual signals for range extensions
-
This method transforms prediction error signals into those in a adaptively computed colour space. A transformation matrix is derived from pixel domain for each coding unit. The colour-space transformation is applied to prediction error of both intra and inter mode. No additional signalling is necessary.
Test 2.2: Fixed colour space transforms (JCTVC-M0048)
-
Matrix approximation of the transforms:
-
YCbCr = [0.213, 0.715 0.072; -0.117, -0.394, 0.511; 0.511, -0.464, -0.047] (Rec. 709)
-
YCoCg = [1/4 1/2 1/4; 1/2 0 -1/2; -1/4 1/2 -1/4]
-
YFbFr = [5/16 3/8 5/16; -1/2 1 -1/2;1 0 -1]
-
FbFr2 = [3/16 5/8 3/16; -1/2 1 -1/2;1 0 -1]
-
YCoCg, YFbFr and YFbFr2 are implemented with the lifting scheme
-
Test 3: Combination of methods (JCTVC-M0049)
-
Test 1.1 + Test 1.2
-
Test 1.1 + Test 1.2 + Test 2.2 (YFbFr2)
-
Test 1.1 + Test 1.2 + Test 2.1 + Test 2.2
-
YCbCr
-
YFbFr2
Test 1.1 LM Chroma first variant:
|
|
All Intra HE Main-tier
|
All Intra HE High-tier
|
All Intra HE Super-High-tier
|
|
|
Y
|
U
|
V
|
Y
|
U
|
V
|
Y
|
U
|
V
|
Test 1.1
|
YCbCr 4:4:4
|
-2.1%
|
-7.5%
|
-7.0%
|
-2.3%
|
-4.8%
|
-6.3%
|
-2.3%
|
-3.2%
|
-4.7%
|
|
YCbCr 4:2:2
|
-1.8%
|
-5.6%
|
-4.4%
|
-1.5%
|
-2.9%
|
-2.8%
|
-1.1%
|
-1.7%
|
-1.6%
|
Test 3: Additive gain of LM and RM
|
|
All Intra HE Main-tier
|
All Intra HE High-tier
|
All Intra HE Super-High-tier
|
|
|
Y
|
U
|
V
|
Y
|
U
|
V
|
Y
|
U
|
V
|
Test 3.1
|
YCbCr 4:4:4
|
-2.4%
|
-9.5%
|
-9.7%
|
-2.7%
|
-6.2%
|
-8.5%
|
-2.7%
|
-4.3%
|
-6.1%
|
|
YCbCr 4:2:2
|
-1.9%
|
-7.0%
|
-6.1%
|
-1.6%
|
-3.6%
|
-3.7%
|
-1.1%
|
-2.1%
|
-2.1%
|
Discussion:
-
Was complexity analyzed?
-
Among 1.1 and 1.2 tests (which only apply to intra), the first tested variant performed the best (YUV) domain. The 2nd variant of 1.1 was proposed as a lower complexity scheme than the first variant, but had some boundary sample memory increase and lower performance and had a division problem for the 4:2:2. The 1.2 scheme is also somewhat complex as compared to 1.1. It was remarked that the use of different reference boundary samples in the LM chroma scheme is a complexity issue, and this had not changed. The gain for AI was about 2% for luma and 5% for chroma (about 3/4 of that for 4:2:2, more for 4:4:4).
-
RM scheme does not significantly help beyond LM, so not particularly interesting as a combination.
-
The LM scheme has decoder inference of the scaling parameter estimated at the TU level based on boundary sample values.
-
The CU-adaptive colour transform scheme tested 2.1 has relatively little benefit relative to LM for AI YUV 4:4:4 and does not work properly (as tested) for 4:2:2.
-
It was remarked that the test 2 colour conversion may have had some issue for YCbCr conversion equation calculations.
-
In test 2, per component results varied. It was remarked that the relative quantization fidelity assigned to the three components will affect the results, and the different colour transforms each produce different energy and correlation characteristics. The relative importance of the three components seemed difficult to ascertain. The BD BR curves were suggested to potentially have some overlap-region measurement reliability issues. Ideally it would be desirable to align the R-D curves for two of the three components to examine the effect on the third. Overall, the results seemed very difficult to analyze. Different colour transforms would each have different characteristics (although many would probably be better than RGB). It was remarked that the bit allocation generally assigned about 60–65% of the bits to luma (very similarly) for all of the tested cases, except when coding in RGB – for which each component got roughly the same number of bits.
-
Out of the tested schemes, it seems that LM chroma variant 1 is the most promising. It only applies to intra. However, the luma/chroma cross-pipeline dependencies and reference sample set difference remain troubling.
Plan: Further study in CE for LM chroma, including testing alternative reference sample set and ultra-high bit rates.
4.1.2RCE1 primary contributions
JCTVC-M0048 RCE1: Adaptive Color Transforms for Range Extensions [W. Dai, M. Krishnan, P. Topiwala (FastVDO)]
JCTVC-M0384 RCE1: Crosscheck of JCTVC-M0048 on adaptive color transforms for range extensions [C.-W. Hsu, Y.-W. Huang (MediaTek)] [late]
JCTVC-M0049 RCE1: Combined Chroma Tools For Range Extensions [P. Topiwala (FastVDO)], J. Kim (LGE), K. Kawamura (KDDI)]
JCTVC-M0097 RCE1: The performance of extended chroma mode for non 4:2:0 format [J. Kim (??)]
JCTVC-M0387 RCE1 Test 2.1: Cross-check of ‘In-loop color-space transformation of residual signals for range extensions’ by KDDI [C. Rosewarne, M. Maeda (Canon)] [late]
JCTVC-M0410 RCE1: Results of Inter-plane intra coding of residual signals [K. Kawamura, T. Yoshino, S. Naito (KDDI)] [late]
JCTVC-M0411 RCE1: Results of in-loop color-space transformation of residual signals [K. Kawamura, T. Yoshino, S. Naito (KDDI)] [late]
4.1.3RCE1 cross checks
JCTVC-M0084 RCE1: Cross-verification of Test 1.1 - LM mode [S. Lee, C. Kim (Samsung)]
JCTVC-M0085 RCE1: Cross-verification of Test 1.2 - Inter-plane intra coding of residual signals (RM) [S. Lee, C. Kim (Samsung)]
JCTVC-M0367 RCE1: Cross-check of tests 3.1, 3.2 and 3.3 [J. Sole, W.-S. Kim (Qualcomm)] [late]
Dostları ilə paylaş: |
