International organisation for standardisation organisation internationale de normalisation
Non-CE technical contributions 3D standard development
Download 8.47 Mb.
|
Non-CE technical contributions
15.0.1.1.1.1.1.1.187JCT3V-E0137 AHG 4: Post Processing in interlace coding mode for MVC+D [C.-C. Lin, F.-C. Chen, C.-H. Hung (ITRI)] This document describes the issues and corresponding solutions for the non-mandatory post processing tools in ATM software in interlace coding mode for MVC+D. Before the fourth JCT-3V meeting, the interlace coding tool had been integrated into ATM and was released as ATM7.2. The previous efforts on interlace integration were mostly concentrated on the mandatory tools. However, we find that there are anomalies in the two post processing tools in interlace coding mode. In addition, the post dilation filter that is not a normative part of MVC+D specification text is automatically applied in the current ATM. For conformance testing, ATM needs an option to enable/disable the depth dilation filter. Three actions are suggested:
Decision: Adopt JCT3V-E0137 (SW and ATM text). The proponents will provide the necessary changes of ATM text to the text editors. 15.0.1.1.1.1.1.1.188JCT3V-E0131 AHG 4: Cross check of Post Processing in interlace coding mode for MVC+D (JCT3V-E0137) [C.-F. Chen, G. G. Lee, B.-S. Li (NCKU)] [late]
15.0.1.1.1.1.1.1.189JCT3V-E0245 3D-AVC: Report on coding performance of the depth coding tools for the 3D-AVC [Gun Bang, Gwang soon Lee, NamHo Hur (ETRI), Kyung Yong Kim, Jung Hyun Kim, Young Su Heo, Gwang Hoon Park (KHU), Ying Chen, Xin Zhao, Li Zhang, Marta Karczewicz (Qualcomm)] In this contribution, all depth coding tools have been tested under the CTC and other configurations. To evaluate the depth coding tools such as DISP, PSIP and IVMP, we prepare additionally full resolution depth case and the texture first coding order configuration. In addition, DISP uses just only AVC deblocking without any changing process and VSO impact are also considered in this contribution. In the result of DISP with AVC deblocking and VSP off case, the efficient coding gain and the complexity reduction could be investigated under CTC configuration and the texture first coding order with MB-level NBDV The result of this contribution shows that the DISP following AVC deblocking and IVMP for depth coding can improve coding gain and the depth coding tools are very useful especially for the full resolution depth coding.
Only test 1 and test 3 are CTC (half resolution depth)
The columns “depth coding” measure the PSNR of depth maps, which is usually not done in our experiments The depth coding tools which were in the PDAM were removed upon request of two NBs. The results given for CTC are not significantly different from the results we had at the last meeting which were used to make the decision on identical depth coding for MVC+D and 3D-AVC (benefit of specific depth coding tools would be only around 1.5% on average). Test 2 and test 4 are with full-resolution depth maps, which if used without specific depth coding tools would end up with approx. 2-3% loss (reported by Dong who used same QP for texture and depth), such that the reported gains of 5-5.5% would also be 2-3% at most when compared to CTC (but full resolution depth map coding is also more complex). The gain reported in test 5-8 (texture first coding), when compared to CTC would only be around 1.5%. The compatibility of depth map decoding between MVC+D and 3D-AVC is advantageous, as it can be done with identical devices, and transcoding of MVC+D streams to 3D-AVC streams is much simpler, because it only needs to be done for the dependent views’ texture (which give most of the compression gain already). No action on the request made in E0245. 15.0.1.1.1.1.1.1.190JCT3V-E0260 3D-AVC: Cross-check of the report on coding performance of the depth coding tools for the 3D-AVC (JCT3V-E0245) [S. Shimizu, S. Sugimoto (NTT)] [late] 15.0.1.1.1.1.1.1.191JCT3V-E0268 3D-AVC: Crosscheck on coding performance of the depth coding tools for the 3D-AVC (JCT3V-E0245) [X. Zheng, Y. Lin (Hisilicon)] [late]
15.0.1.1.1.1.1.1.192JCT3V-E0230 AHG13: MFC Conformance Testing [P. Yin, T. Lu, H. Ganapathy, T. Chen, W. Husak (Dolby), D. Tian (MERL)] This document provides information about MFC conformance Testing. It has two attachments. The first attachment is the draft text of MFC conformance testing. The second attachment is a set of candidate bitstreams for MFC conformance testing. 6 bitstreams (17 pictures, 1 GOP) were used to test the basic functionalities (RPU modes, MFC format, interlaced/non-interlaced). 15.0.1.1.1.1.1.1.193JCT3V-E0231 AHG13: MFC Reference Software Update [P. Yin, T. Lu, H. Ganapathy, T. Chen, W. Husak (Dolby)] This contribution provides an update on the MFC reference software source code and software manual. The software includes an implementation of Frame Packing Arrangement SEI message based on the MFC specification and fixes several issues. Draft 1 of MFC conformance (PDAM as MPEG doc). Draft 1 of MFC reference software (PDAM as MPEG doc). MFC Test Model v2 (no change, but version 1 included the old ref soft).
15.0.1.1.1.1.1.1.194JCT3V-E0100 JCT-3V AHG 3: Editorial improvements on MV-HEVC Draft Text 4 [G. Tech, K. Wegner, Y. Chen, M. Hannuksela, J. Boyce] No need for detailed review, was included in AHG report. Ballot results (MPEG M30005) were reviewed. On FI-2 (depth map type in auxiliary pictures), this needs coordination with RExt, see also E0102. Gerhard will prepare disposition of comments. One comment requests for definition of levels, which does not exist so far. A Breakout meeting was held (coordinated by A. Vetro, M. Hannuksela) Thu 9:00. The BoG will produce a draft text with the level definitions for MV-HEVC draft, to be uploaded and approved Friday morning. 15.0.1.1.1.1.1.1.195JCT3V-E0310 BoG report on draft text for level definitions in MV-HEVC stereo profile [A. Vetro, M. M. Hannuksela, Y. Chen] The document was reviewed Friday morning. The suggested specification is in many regards only applicable to the stereo case (in terms of level and related DPB size constraints). Therefore, it is asserted as being a first step requiring more careful thinking to make it more general and extensible to multi-view and 3D cases. Decision: Adopt. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||