Video Compression

VideoNerd

HEVC supports non-uniform (or arbitrary) tiling.

Therefore encoder should adapt tiling grid in order to minimize the decoding time. Bitsizes of all tiles are expected to be nearly equal each to other. Otherwise, the decoding process can be imbalanced ( if multiple threads are applied to decode tiles simultaneously then small sized tiles are finished earlier and a heavy tile would get a bottleneck in performance).  

Note: in addition to parallelization, there  are applications of tiles in 360-videos. In the 360 video a user always sees a part of frame (it’s called viewport), therefore either a decoder could choose and then decode relevant tiles or only relevant tiles are transmitted by the server. The latter variant is better since redundant tiles are not signaled and the bandwidth is not impacted.

In both solutions motion vectors should be confined to reference within tiles, since a part of reference temporal data can be non-available.

8 Responses

  1. Thanks for the sensible critique. Me & my neighbor were just preparing to do a little research about this. We got a grab a book from our local library but I think I learned more from this post. I am very glad to see such fantastic info being shared freely out there.

Leave a Reply

Your email address will not be published. Required fields are marked *