Open dankamongmen opened 3 years ago
Remember, moving to support more than 256 colors means we'll have to move to a two-byte lookup in extraction.
having thought about this some more:
but do not yet take the color count beyond 256, even if available. we need a faster quantizer to take good advantage of the extra space. on the other hand, the current quantizer absolutely requires at least 64 color registers, rejecting fewer in sixel_blit()
, so getting that pumped up is kinda important.
following the work in #1761 and #1469, we attempt to set the number of color registers to 256 before reading the number of color registers. we maybe ought try to set it to 64 first, in case it supports 64 but not 256 (remember, 64 is the minimum with our current algorithm)? we're not yet doing anything with sixel maxes.
how is this distinct from #2141?
1512 mentions that we might as well take the number of color registers, and maximum sixel size, up to the maximum reported supported values. note that there's a speed cost if we end up using all those extra color registers.
is there any other reason why we wouldn't want to do this? i feel that there's likely lost knowledge and esoteric arcana in this area.