Hi Pgyuri,
I was thinking 128K only, using double buffering and running at 25fps.
Also, I've been thinking about using "compiled sprites", which are the fastest way to draw sprites on Z80. So you have a routine per sprite which directly loads values from constants and writes them to VRAM.
I have also been thinking about a cunning way of reducing CPU load for removing old sprites. You basically keep dirty block regions and only replace those regions which aren't going to be overdrawn by sprites again.
However I've not done a lot of Z80 as I've always been a 68000 guy, so it'll be interesting to see how slow this all is!
And as a final resort, when I get my Enterprise on and FPGA running, I'll add some hardware sprites through the EC0-3 inputs.
Still amazing to see what the Enterprise was capable of displaying in terms of graphics. What it could move around the screen is another thing, though...
James