Search Results for "JVET-R0293"

Found 5 document(s)

Search documents

Use number, keyword, author, or MPEG number. Filter by meeting when needed.

18th Meeting: Teleconference, April 2020 2020-04-17 15:03
Abstract
It is asserted that the some of the current 6-tap interpolation filters for affine motion compensation have significant phase misalignments compared to the expected phases of the interpolation filters. This contribution proposes modifications of the filters for the fractional positions 4/16, 5/16, 6/16, 10/16, 11/16 and 12/16, and it is claimed that this fixes the phase misalignments of the current filters.
JVET-R0293 Fixes for 6-tap interpolation filtering for affine motion compensation [K. Andersson, R. Yu, Z. Zhang, J. Ström (Ericsson)]

It is asserted that the some of the current 6-tap interpolation filters for affine motion compensation have significant phase misalignments compared to the expected phases of the interpolation filters. This contribution proposes modifications of the filters for the fractional positions 4/16, 5/16, 6/16, 10/16, 11/16 and 12/16, and it is claimed that this fixes the phase misalignments of the current filters.

The modifications were reported to be tested under VTM-8.0 CTC. The impact on coding efficiency is reported to be -0.01%/-0.09% for RA/LDB. The modifications were reported to also be tested for the non-CTC) affine test set which was used to determine the performance of proposals in the past affine CEs. The impact on coding efficiency on that test set is reported to be -0.20% for RA.Presented Fri 17 April

It is mentioned that phase is less important in motion comp, amplitude/energy preservation more important. It is also mentioned that even the 8-tap filters are not optimum in phase. New results in v4 include the energy gain, which show that it is slightly lower for 6/16 filter than for the current filter, whereas for the other two (4/16 and 5/16) it is slightly higher.

It was also asked if the visual effect of the filters was investigated. The proponent says that no difference was visible.

No action (see further notes under JVET-R0292).

Decisions
No action (see further notes under JVET-R0292).
Citation
18th Meeting: Teleconference, April 2020 2020-04-17 19:22
Abstract
Contribution JVET-R0293 proposes to modify six of the 15 6-tap interpolation filters that are used for subpel interpolation filtering in affine inter predicted blocks. New filters are proposed for the fractional positions 4/16, 5/16, 6/16, 10/16, 11/16, and 12/16. The software provided by the proponents has been verified to match the text provided by the proponents. Coding efficiency results using the VTM-8.0 SDR Common Test Conditions (CTC) have also been confirmed. Additionally, the frequency responses of the proposed interpolation filters have been computed and plotted by the crosschecker. The resulting graphs match those provided in JVET-R0293.
JVET-R0475 Crosscheck of JVET-R0293 (Fixes for 6-tap interpolation filtering for affine motion compensation) [M. Winken (HHI)] [late]
References:
JVET-R0293
Decisions
JVET-R0475 Crosscheck of JVET-R0293 (Fixes for 6-tap interpolation filtering for affine motion compensation) [M. Winken (HHI)] [late]
Citation
18th Meeting: Teleconference, April 2020 2020-05-17 19:13
Abstract
This document provides general summary information about typical quality measurement practices for video coding experiments in recent work in MPEG and VCEG and their JVET and JCT-VC joint teams. In particular, it describes the use of Bjøntegaard Delta rate (BD-rate) measurements. It aims at providing a concept-level overview of recent practices and to provide some references to technical papers that describe further details. It provides comments on why some of the choices were made and points at situations where caution must be taken when interpreting the results. Revision marks are shown relative to the January 2020 output document, JVET-Q2016-v4.
JVET-R2016 Summary information on BD-rate experiment evaluation practices [K. Andersson, F. Bossen, J.-R. Ohm, A. Segall, R. Sjöberg, J. Ström, G. J. Sullivan, A. Tourapis] (2020-05-15)

Minor refinements of the description were planned.

Future meeting plans, expressions of thanks, and closing of the meeting

Future meeting plans were established according to the following guidelines:

  • Meeting under ITU-T SG 16 auspices when it meets (ordinarily starting meetings on the Tuesday of the first week and closing it on the Wednesday of the second week of the SG 16 meeting – a total of 9 meeting days), and
  • Otherwise meeting under ISO/IEC JTC 1/SC 29/WG 11 auspices when it meets (ordinarily starting meetings on the Wednesday prior to such meetings and closing it at lunchtime on the last day of the WG 11 meeting – a total of 9.5 meeting days).

In cases where an exceptionally high workload is expected for a meeting, an earlier starting date may be defined.

Some specific future meeting plans (to be confirmed) were established as follows:

  • Mon. 22 June – Wed. 1 July 2020, 19th meeting under ITU-T auspices conducted by teleconference.
  • Wed. 7 – Fri. 16 October 2020, 20th meeting under WG 11 auspices in Rennes, FR.
  • Wed. 6 – Fri. 15 January 2021, 21st meeting under WG 11 auspices in Capetown, ZA.
  • Tue. 20 – Wed. 28 April 2021, 22nd meeting under ITU-T auspices in Geneva, CH.

The agreed document deadline for the 19th JVET meeting was planned to be Wednesday 10 June 2020 (earlier than usual).

Vittorio Baroncini, Andrew Segall, Mathias Wien, and Yan Ye were thanked for their efforts in further...

Decisions
Tianyang Zhou (Sharp)
Citation
18th Meeting: Teleconference, April 2020 2020-04-18 13:00

Abstract not available in document

JVET-R0340 Agenda and report of the category 2 AHG pre-meeting of the 18th JVET meeting [J.-R. Ohm, B. Bross, A. Segall, Y. Ye] The category 2 AHG pre-meeting of the 18th JVET meeting, for addressing coding tool topics, mostly in the scopes of AHGs 2, 3, 6, 11, 14, 15 and 16, held teleconference meeting sessions on April 9 and April 14. Approximately 227 people attended the AHG pre-meeting, and approximately 45 input documents were discussed. The following recommendations were made to the JVET: JVET-R0223 (Ed.) Editorial improvement of DMVR process left to editor. JVET-R0134 (BF text) Mismatch related to deblocking of subblock motion edges JVET-R0168 (BF text) Issue on bS derivation of deblocking filter (check interaction with JVET-R0228) JVET-R0228 (BF text) Bugfix to deblocking filter boundary strength setting for palette JVET-R0208 (cleanup/text+software) Rounding correction for ALF v...
Decisions
JVET-R0083, JVET-R0116 Interaction of DQ, SDH and TSRC (also JVET-R0141, JVET-R0144, JVET-R0154, JVET-R0325)
Citation
18th Meeting: Teleconference, April 2020 2020-04-17 15:03
Abstract
It is asserted that the current 4-tap interpolation filters which are used both for intra angular prediction and chroma motion compensation have significant phase misalignments compared to the ideal phases of the interpolation filters. For ten of the interpolation filters the phase misalignments are as large as the expected phase differences between two interpolation filters of adjacent fractional positions. This contribution proposes 4-tap interpolation filters that fixes the phase misalignments of the current 4-tap interpolation filters.
JVET-R0292 Fixes for 4-tap interpolation filtering [K. Andersson, R. Yu, Z. Zhang, J. Ström (Ericsson)] Only one aspect related to inter: 4-tap interpolation filter for chroma motion compensation 4-tap interpolation filter for intra angular prediction It is asserted that the current 4-tap interpolation filters which are used both for intra angular prediction and chroma motion compensation have significant phase misalignments compared to the ideal phases of the interpolation filters. For ten of the interpolation filters the phase misalignments are as large as the expected phase differences between two interpolation filters of adjacent fractional positions. This contribution proposes 4-tap interpolation filters that fixes the phase misalignments of the current 4-tap interpolation filters. The fixes were reported to be tested under VTM-8.0 CTC. The impact on coding efficiency is reported fo...
Decisions
No action was taken on this.
Citation
New Search