JVET-R0307 AHG8/AHG9: Positioning information SEI message of output independent layers [E. Thomas (TNO)]
This contribution presents a multi-party video conferencing use case where a multi-layer bitstream with independent layers can be used. Each participant views can constitute an independent layer of a single multi-layer bitstream which forms a mosaic or also called grid view.
This contribution proposes defining a new SEI message to signal the positioning information of each output layer for a given output layer set.
In addition, the associated contribution JVET-R0308 provides the software implementation in VTM corresponding to the presented scenario that is:
- Implementation of a multi-layer decoding support
- Composition into a single picture of independent output layers of the same output layer set
It was commented that sample aspect ratio could be important to consider.
It was commented that various compositing syntax methods could conceivably be used.
The compositing scheme could have some rectangles not occupied.
There were several previous contributions relating to this concept.
There was some interest expressed in the concept of a compositing layout signalled in an SEI message.
One justification given for the syntax approach was the ability to remove layers without changing the relative positioning of the remaining layers.
The approach was described as using a 4-neighbour connected graph.
It was commented that video mixing is sometimes conducted but typically in a somewhat different way.was suggested to consider this for inclusion when a draft of SEI messages is developed.
Further study was encouraged, including study of the available software. The changes to support this in the software was said to be relatively minimal. The proponent said they could also provide example bitstreams.