I've decided that I'll give my ideas to whoever wants them, as long as they make it into open-source freeware.

I had the idea a long time ago to alter the current palette-based bitmap compression by localizing a few smaller palettes. Instead of being limited to using 256 colors in a single picture, you could use as many as you wanted, with the only limit being how many unique colors could fit into a certain block.

Maybe that wasn't clear. The standard in Windows is a single 8-bit palette for a bitmap that could stretch out to 16,000² pixels or more.

Do you see the problem there?
Of course there's PNG, but it gets much more efficient with an 8-bit palette.

I thought that a block of pixels could be indexed to multiple small palettes, and these palettes would be made by some sort of algorithm, choosing based on the simplest blocks or the most complicated. I guess the math on how to make the optimal palette is up to you, but I figured that a 16² block should not be the smallest block if you only planned on allowing 8-bit indexing.

Of course, anybody really serious about this compression should accommodate for 1, 2, 4-bit and possibly 6, 12-bit indexing.

Thanks for reading this far. I do not apologize for not posting more.