no, I'm using just dispgraphrr, and two sets of 4-level grayscale. Unfortunately, it behaves similarly to 3-level when done like that, so with 3, 2, 1 and 0 representing each sprite's 4 level value, 3 over 1 is the same as 1 over 3, and so on, giving only 9 values from the 16 pixels in my test pattern.

If I had the evolving pattern from the backbuffer also on the front buffer, this would be true 16-level.

I'm not sure how effective combining all three dispgraphs would be, since they basically build on each other.

Or you can set apart a 768-byte portion of your program :S

Oh, right.

Just thought of something: Wouldn't it technically take less RAM to just use labels like GDB1 inside the program? Because initializing a value in L₁ takes up 6 bytes or something, whereas initializing a byte to a label is just that one byte.

You'll need your own (ASM) routine

Well, 3 level and 4 level include white, so you're probably good with 9.

Good luck with your entry. I'd like to see 9-lvl grayscale in Axe

whoa!  what game is that?!
guess, TI-Boy?

This was a Zelda mockup by Zera a while ago, not an actual program. The above demonstrates the grayscale difference between each calc. If you pay attention to the grays in the 83+ one, you'll understand why I don't recommend anything above 6 lv on that calc.

well that'S good to hear. I can't wait to see it in action.