Back to Search Document details
17th Meeting: Brussels, January 2020 2020-01-11 12:39
Mandatory film grain
Authors: Rickard Sjöberg Ericsson D.Saffar Ericsson Martin Pettersson Ericsson Mitra Damghanian Ericsson
Abstract
This contribution proposes to add normative film grain generation to VVC, within conformance. When enabled, the generated film grain is proposed to be applied to the decoded picture before output, specified in Annex C, similar to how picture cropping is handled.
Meeting Notes
JVET-Q0424 Mandatory film grain [R. Sjöberg, D. Saffar, M. Pettersson, M. Damghanian (Ericsson)]

This contribution was discussed in the plenary of Sunday 12 January (chaired by GJS & JRO).

JVET-Q0424 and JVET-Q0533 propose a new tool for consideration as a mandatory required feature of profiles (without evidence or complete text). The proposal is different from what was used for an SEI message in AVC and HEVC.

It was commented that this is interesting but does not seem feasible for inclusion in v1 of VVC.

Further study in an AHG was encouraged, but not planned as a potential mandatory feature of v1 of VVC.

The AHG study should include development of evidence, text and software for an SEI message approach.

Transcoding was a suggested relevant use case.

It was agreed that we do not necessarily need to remove the existing SEI method (which was carried over from HEVC and from AVC to HEVC) from the text or to require a comparison of the proposed new method with the prior two models, since software for the prior models is not available. However, having software for the existing draft models would be highly desirable.

References:
JVET-P2007 JVET-Q0041
Cited:
JVET-Q0533 JVET-Q0614 JVET-R0017 JVET-R0384 JVET-Z0132
PATENTS:
WO/2022/220724 0.34 20240205460 0.33 3215231 0.32 2022257959 0.31 WO/2024/213679 0.30 20250056064 0.28 WO/2023/110642 0.27 WO/2022/194866 0.26
ARCHIVE v2 :
JVET-Q0424-presentation.pdf JVET-Q0424-v2.docx
Decisions
It was agreed that we do not necessarily need to remove the existing SEI method (which was carried over from HEVC and from AVC to HEVC) from the text or to require a comparison of the proposed new method with the prior two models, since software for the prior models is not available. However, having software for the existing draft models would be highly desirable.
Citation