The following issues were initially identified to be resolved during the meeting:
-
Status and timeline (see notes under closing plenary)
-
Maturity of WD (see notes by the end of this section)
-
NB comments, inputs below were discussed Tue 10:00
The following input documents were reviewed and recordings of discussions were made in the BoG (chaired by Mohammed Raad for technical discussions).
14.1.1.1.1.1.1.1.72m30772 USNB Comment on “Video Coding for Browsers” [A. G. Tescher for USNB]
See disposition in N13949. The announced support in the VCB development is appreciated.
14.1.1.1.1.1.1.1.73m31561 JNB comment on the activities from the CfP for Internet Video Coding Technologies
See disposition in N13949 and further discussion during the Wednesday plenary. More information will be acquired about the benefits of the different codecs, including a visual (subjective quality) comparison to be performed in San José.
14.1.1.1.1.1.1.1.74m31666 Finnish NB comment on the VFB activity [J. Ridge on behalf of FI NB]
Item 1 not business of video SG.
Item 2:
-
Google has responded on resolution 15.1.7
-
Situation is different than IVC, where evidence of “prior art” is taken as a clue for type-1. It is obvious that IPR exists in VCB (similar as it is in WebVC).
-
Further consideration requires advice by ITTF
From the work performed in the video group, a decision on whether progressing to CD would be based on maturity of the technical description in the draft text to meet the quality requirements of our standards. In particular, there should be no missing elements to understand the structure of the bitstream, its semantics and the operation of the decoder from the text specification.
Item 3:
-
Will the proposed ‘VCB’ standard contain only a decoder specification, or is it proposed to standardize aspects of the encoder as well? Answer: NO (this is the only point that can be answered by MPEG)
-
Does the proponent’s offering of their own IPR on a “Type 1” basis extend to the ‘VCB’ encoder technology?
-
Does the proponent’s statement regarding MPEG-LA licenses extend to the ‘VCB’ encoder technology?
-
Referring to MPEG resolution 15.1.7 of the 105th meeting, does the list provided in document m31231 include encoder technologies?
During the discussion, the Google representative indicated that the answer is YES on bullet points 2-4.
14.1.1.1.1.1.1.1.75m31231 Response to N13648 resolution 15.1.7 (regarding knowledge of IPR relevant to VCB) [Harald Alvestrand, Mohamad Raad]
M31231 responds to resolution 15.1.7.
14.1.1.1.1.1.1.1.76m31255 A comment on the VCB project [Harald Alvestrand, Maire Reavy, Timothy Terriberry, Mohamad Raad]
Contribution noted.
Several discussion sessions were were in the video plenary and BoG regarding the maturity and improvement of the WD. Intense work happened offline between the editors on improving the draft text according to standard quality of MPEG standards such that in the new version
-
the sequence of chapters is basically OK
-
Software and (non-normative) encoder included as annexes
-
arithmetic entropy decoder included now, but not cleanly described as pure decoding process
Several experts also studied the text during the week and pointed out various remaining weaknesses, such as
-
unprecise expressions of computation accuracy
-
vague description of (some) processes
-
many parts describe implementation rather than input –> output of decoder processes
-
inheritance from "specification per software" obvious
-
too much verbosity, unnecessary duplications
It was concluded that the current text is not mature enough for a promotion to CD.
It was also generally questioned whether it would be possible to implement a VCB device on the basis of the text specification, and whether the current text is matching with the software. This also requires further assessment
Reconfigurable Media Coding – Video related
Discussed during Wednesday plenary –
The breakout group on reconfigurable media coding had met under the 3G subgroup. Items related to RVC were reported during the Wednesday plenary of the video group and confirmed as follows:
Document No.
|
Title
|
Authors
|
Notes & Recommendations
|
m31291
|
Generic parser FU implementation for MPEG-4 SP RSM
|
Hyungyu Kim, Seungwook Lee, Bon Ki Koo, Euee S. Jang
|
Promising contribution for the generic parser FU, a demonstration has been show cased with the MPEG-4 SP decoder from the VTL. The generic parser has been used instead of the static parser already in use in the VTL.
BSD of MPEG4-SP decoder has been translated by hand into an internal representation. further work will be investigated to automatically translate this BSD for the next meeting cycle.
|
m31290
|
Proposed revision of RVC-BSDL for the bitstream parser instantiation
|
Hyungyu Kim, Seungwook Lee, Bon Ki Koo, Euee S. Jang
|
Slight changes to the RVC-BSDL are proposed. Due to the tool to generate the RVC-BSDL into the intermediate representation used by the AlgoSynP that is not yet ready, this contribution is delayed to the next meeting.
|
It was concluded that further investigation in CE is necessary for one more meeting cycle (see CE description N13930). There is also more need to clarify whether this only requires amendments to 23001-4 and 23002-4, or also corrigendum actions.
m31403
|
On using tiles for a parallel HEVC decoder description (post implementation)
|
Khaled Jerbi, Damien De Saint Jorre, Mickaël Raulet, Marco Mattavelli
|
This contribution presents a parallel video HEVC decoder based on the FU Network description of HEVC. It extends it with only one FU and reuse all existing FU with only minor modifications to the Parser.
|
m31400
|
Linking SSE-optimized functions within a RVC-CAL program
|
Khaled Jerbi, Mickaël Raulet
|
This contribution provides informative ways to link already existing and low level optimized functions into an RVC CAL program.
|
It was concluded that the development of HEVC related tools is mature enough to proceed to DAM. Editors were asked to prepare the DoC on 23002-4/PDAM2 and 23002-4/DAM2 for approval on Thursday.
No missing parts are known in text, cases where HEVC conformance streams do not pass are software implementation specific.
14.1.1.1.1.1.1.1.77m31290 Proposed revision of RVC-BSDL for the bitstream parser instantiation [Hyungyu Kim, Seungwook Lee, Bon Ki Koo, Euee S. Jang]
14.1.1.1.1.1.1.1.78m31291 Generic parser FU implementation for MPEG-4 SP RSM [Hyungyu Kim, Seungwook Lee, Bon Ki Koo, Euee S. Jang]
14.1.1.1.1.1.1.1.79m31400 Linking SSE-optimized functions within a RVC-CAL program [Khaled Jerbi, Mickaël Raulet]
14.1.1.1.1.1.1.1.80m31403 On using tiles for a parallel HEVC decoder description (post implementation) [Khaled Jerbi, Damien De Saint Jorre, Mickaël Raulet, Marco Mattavelli]
Dostları ilə paylaş: |