International organisation for standardisation organisation internationale de normalisation

5.5Software & Testing

m23718	Dmytro Rusanovskyy	3DV-ATM anchor bitstreams production under the common test conditions
m23719	Dmytro Rusanovskyy	Validation of the 3DV-ATM software with Nokia's response to the MPEG 3DV CfP
m23783	Krzysztof Wegner, Jakub Siast, Jacek Konieczny, Olgierd Stankiewicz, Marek Domanski	Poznan University of Technology tools for 3DV coding integrated into 3D-HTM

m23783 (Poznan) reported that a number of tools have been integrated into the 3D-HTM software, including view synthesis based on inter-view disocclusion prediction, depth-based motion prediction, adjustment of QPs of texture data based on depth data, non-linear depth representation. A number of these tools are specifically designed to improve subjective quality. The software is available in the MPEG SVN.

m23718 (Nokia) indicated that the 3D-ATM software had been released with all tools from Nokia and Qualcomm as planned. There was one tool disabled. Two coding configurations were included: one that is MVC compatible with DRWP on, and another that is AVC compatible with all tools on. An Excel template was provided for reporting the coding efficiency of each reconstructed component, as well as the total bit rate versus synthesis quality. It was noted that some updates to common test conditions (CTC) should be made to reflect this reporting. Anchor bitstreams were generated for all cases. It was agreed that this should be distributed among several participants. It was concluded that better cross-checking and verification is needed across different platforms.

The following issues were discussed and agreed:

• 
  It was suggest to perform encoding C3 and extract C2 results, and to specify this as part of a CTC document. Some issues were noted, e.g. requiring bitstream extraction software.
  
• 
  Down/up-sampling methods for CTC anchor shall be explicitly stated.
  
• 
  Some version of input control data shall be introduced and stated in the CTC document.
  
• 
  RD result reporting could be based on suggested template and metrics; we need further discussion on measuring.
  
• 
  To resolve an issue with mismatch of view synthesis, we consider using the HHI implementation of view synthesis software.
  
• 
  We need to develop an "experimental starter kit" including all configurations, control for input data, reporting template, etc.
  

m23719 (Nokia) reported on a validation of 3D-ATM software against an original proposal from Nokia. Minor deviations were found, but nothing significant.