Jukka Andberg wrote: > Working solution seems to be creating the shadowfb with 8bpp and > depth 4, and then handling conversion from 8bpp pixels to 4 > bitplanes in the shadow update function. > > For the conversion it needs a new shadow update function, which is > pretty similar to existing shadowUpdateAfb4, just reading 8-bit > input pixels instead of 4-bit pixels. > > Patch attached. Any thoughts? I think this could work. Nice work! Maybe it makes sense to optimize at least the inner loop of the shadowUpdateAfb4x8() routine into assembler, once it works. -- Frank Wille