Following up on that: if it would be useful, I can cut the part of that stream where data partitioning is actually used down to a FATE-sized excerpt and submit it to the samples repo. The file is about 80 MB as it stands, so it needs trimming either way.
That excerpt could sit alongside the JM-encoded streams the series currently uses for its tests, or replace them wherever it covers the same case. I would want to keep a constrained_intra_pred_flag stream in some form, since that is the one that exercises the 9.2.1 nC derivation, but if the real stream covers that too then the artificial one is redundant. Happy to do that work if it would make the series more palatable. _______________________________________________ ffmpeg-devel mailing list -- [email protected] To unsubscribe send an email to [email protected]
