m32253 Constraint Signalling for HEVC coded Still Images and Image Sequences
Accepted into the still image format with (a) maxrefpic==15 meaning ‘whatever the level allows’ and no optional box() at the end.
-
Thank you for the nice discussion of the various possibilities. We initially focused on the simple case (e.g. non-tiled, no layering etc.). We do need to consider tiling, scalability, and so on, at some point.
We noted that in option 3, we think that JP2/JPX were designed as “Jpeg2000 formats” and not layered as “still image formats” which initially are used to contain Jpeg2000. We think it would need a lot of work. However, we note that the both the track and meta-box options reference image data by pointer/offset, so it could indeed still be possible to have the image data in a ‘codestream’ box.
It’s clearly true that if sequences with times are needed or useful, a track structure does the job. But people on the web want to load images fast, and have something simple.
We note that there is a need for a format for responsive images, in which one can select an image from a set by choosing based e.g. on DPI. It’s not clear whether this drives us in any direction, however.
Dostları ilə paylaş: |
