[FFmpeg-devel] a64 encoder
Fri Jan 16 21:21:46 CET 2009
>> You should be calculating the set of chars per frame or video and not
>> hardcode them.
> In case of multicolor mode, i calculate an optimized charset of 256
> chars that cover 4000 blocks from the original 4 frames (thus having a
> bit of quality loss). In case of Petscii i make use of the C64 charset
> only (that the c64 has in its ROM furthermore), and not building any
> dynamic charset on my own, so i can forgo on also putting a charset into
> the video itself, and therefore add additional color information.
> However in that case i need all origional characters from the c64 to be
> able to compare them with each block of the picture.
> In case of ecmh mode i also rely on a static charset, but especially
> prepared for maximum quality.
Is there any advantage of Petscii in comparison with ecmh?
> So having several ways for encoding is part of the idea.
Isn't it possible to have a single file format for all the ways of
encoding? Maybe you could even encode in a way to permit files with
non-constant lifetime like the following example:
CHARSET, FRAME, FRAME, CHARSET, FRAME, CHARSET, FRAME, FRAME, FRAME,
FRAME, CHARSET, ...
More information about the ffmpeg-devel