International organisation for standardisation organisation internationale de normalisation

Conformance

Software

1. General

2. AVC related

4.2.1.1.1.1.1.1.117JCT3V-C0069 AHG4: Interlace coding on ATM Encoder [C.-C. Lin, F.-C. Chen (ITRI)]

Was reviewed in BoG (JCT3V-C0236). Confirmed by proponents that it is planned to have the full integration of interlaced tools into ATM software by the next meeting.

This document describes the issues and development status of ATM interlace support for consideration as MVC+D reference software.

The contribution reports on integrating interlace coding on MVC+D software – ATM. The contribution reported problems in ATM which needs to be resolved in order to support interlace coding in ATM.

As current status of integration work, the contribution reported that software is capable of coding all of test sequences as field, however, one sequence is indicating some blocking effect – which the reason is currently investigated.

Only the encoder is worked on currently. The decoder needs to be worked on as future work.

Future work was introduced to finalize the integration – related contribution JCT3V-C0071.

The contribution proposes coding configuration for the HP profile of ATM based on the integaration.

Comments from BoG:

Some participants commented that the proposed coding configuration should be specified in ATM software description – and the MVC+D ATM software description should be a separated document as independent document.

Recommend to create an independent software description for MVC+D which includes proposed coding configuration. ITRI will provide the initial draft.
4.2.1.1.1.1.1.1.118JCT3V-C0036 Report on results of cross-check on coding with JMVC+Depth software in interlace mode [O. Stankiewicz, K. Wegner (Poznan Univ.)]

This report presents results of cross-check of coding with JMVC+D software initially provided by Sony, further modified by ITRI, performed by Poznan University of Technology.

On one platform (Sentos Linux) no encoder crash in cases of interlaced texture and progressive or interlaced depth. On other platforms, only progressive/progressive works.

Decoding is not a problem on any platform.

T. Suzuki reports that a bug was found in the motion estimation part, which may be a likely reason for the crash.
4.2.1.1.1.1.1.1.119JCT3V-C0164 Recommendation on MVC+D reference software [O. Stankiewicz (Poznan Univ. ), K. Wegner (Poznan Univ.), D. Rusanovskyy (Nokia)] [reject]

In this document advantages and disadvantages of available candidates for MVC+D reference software are discussed. It is observed that neither of available candidates (JMVC or 3DV-ATM) are currently fully compliant with general requirements on MPEG standard reference software and both candidates need additional development efforts to comply with it. However, considering design architecture and stability of software packages, it is recommended to utilize 3DV-ATM software as a reference test model for MVC+D specification. The list of recommended changes is provided in m27809/JCT3V-C0069 and m27811/JCT3V-C0071 MPEG documents.

3DV ATM software (JM based) lacks interlace support (ITRI is working on it, according to C0069), and had a (meanwhile fixed) bug in HL syntax of MVC+D.

JMVC-based software would lack the implementation of substantial parts of HL syntax of MVC+D, and has a bug in interlace (which would be resolved if the report under C0036 comes true).

Due to that situation, at this moment there is no way to make a cross-platform check (e.g. JM encoder, JMVC decoder) of bitstreams (not mandatory, but would simplify verification of the development).

3. HEVC related

No contributions noted.

6. Source video test material

4.2.1.1.1.1.1.1.120JCT3V-C0101 AHG11: 3D Test Materials from NICT-3D data set [S. Shimizu, S. Sugimoto, H. Kimata (NTT)]

In this document, a set of stereo or multiview sequences is proposed for usage as test sequences in the development of 3D video coding technologies. These sequences are distributed from NICT for the purpose of research and development activities.

Different sets:

1. 
   3 sequences 4K stereoscopic (captured by Red camera), 3 minutes each with scene cuts
   

No depth map, no camera parameters

Setups with changed baseline and convergence angle, partially rectified.

The indoor scenes are relatively static.

More interesting: Outdoor scenes with high detail, pan, motion of leaves, fade.

Would be interesting to perform encoding of some parts to find out which are challenging

2. 
   3 views + depth, time-of-flight camera (Axi-Vision). Interlaced (also depth)
   

4 sequences, living, office, dartsbar, and dance, several minutes long

The depth maps are noisy, may not be fully correct

3. 
   graphics content, convergent virtual cameras, dynamically changed convergence angle, precise depth maps and camera parameters could be provided. Also moving virtual cameras, high detailed and transparent objects. Motion blur during rendering?
   

Critical issues in license agreement (to be clarified):

• 
  reporting of usage in conferences, fairs
  
• 
  termination of license on notice by owner
  

Further investigation in AHG which sequences are challenging.

4.2.1.1.1.1.1.1.121JCT3V-C0210 3D Holoscopic Video Test Material [Paulo Nunes (IT/ISCTE-IUL), Luis Soares (IT/ISCTE-IUL), Amar Aggoun (Brunel University)] [late]

In the 101^st MPEG meeting in Stockholm various future work items of JCT-3V on 3D standardization were identified, namely, “3D Video Formats beyond depth” where the 3D holoscopic approach can be included as a prospective candidate technology for future glasses-free 3D technology.

In order to contribute to the advance of this prospective multimedia technology, the 3D VIVANT consortium is willing to provide test material to the JCT-3V group in the form of holoscopic video sequences.

Sequences effectively have 14x14 views (arranged as patches in the demonstrated slide examples), but spatial resolution is relatively low.

Sample data could be made available via ftp site (to be announced via reflector).

7.