Per subject, WIP patch for Escape 130. Only partially
working at the
moment: some images are recognizable, but with lots of
multi-colored
artifacts. I'm reasonably confident that I'm parsing the
bitstream
correctly, since with this sort of codec, a small mistake
tends to
lead to complete garbage.
This patch doesn't need a review at this point; I'm well
aware it's a
complete mess. My biggest question at the moment is whether
decoding
to YUV420P is appropriate for this format. Both the wiki
page and the
reverse-engineered spec make some comments about the codec
doing some
parts in RGB, but I'm not sure how significant that is; the
RGB
adjustments described in the spec are probably invisible.
I have a few theories for the cause(s) of the artifacts:
one, there
could be a bug in the spec I'm working from, or in the
tables in the
spec. Two, I could have misinterpreted the spec somehow,
or
incorrectly converted the tables. Three, I made some stupid
mistake
that I just haven't spotted yet. It would be nice if
someone figured
out the cause of the artifacts; I'm not really expecting
anything,
though. There are definitely errors in the Y plane; with
that being
the case, it's hard to tell whether there are also artifacts
in the Cb
and Cr planes.
-Eli
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel mplayerhq.hu
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel
|
