Video annotations are also interesting to me. In my case for tourism/walking videos there are requests for additional infos ontop of the video. On YouTube it's Subtitles yet WebVMT [0] seems an interesting format for non-destructive info-layers. It would be great to have exports to subtitle format as a fallback and possible integration to WebVMT (image / map thumbnail).
Another use case could be cartoons with (localized) speech bubbles.