Closed dwbuiten closed 7 years ago
Setting repeat_pict to -1 was an overly cutesy way to fit the data in the parser context fields available (obviously repeating a frame -1 additional times means it removes the first appearance, right?), and without that constraint it should just set Invisible directly.
Changed as such. It can be merged once someone tests visually + seeking.
This is the groundwork for upcoming VP9 alt-ref and superframe support.
VP9 already has a parser inside FFmpeg, so we can't register our own, like we do for VP8, so to ready the codebase for this change, this commit separates out the VP8 "parser" into FFMS itself.
This needs some indexing and seeking testing visually with VP8 files before it is merged, and a good look-over by @tgoyne. #