SpiceWare's Blog

Is there not enough CPU time to do without hardware collision detection?

OK, I think that looks perfect now, as it gives the jagged appearance without looking too cluttered like before

 

Chris

I like it! It's definitely an improvement.

 

Especially because the huge contrast between the hires dragon and the lowres playfield has been minimized now.

Hey...does this mean there's going to be a "brick walls" option now?

Is there not enough CPU time to do without hardware collision detection?

Nope - I'd have to check all 4 corners for each fireball on every frame they're in a castle row. Right now, if the first fireball ends up having to check all 4 corners to find the brick it collided with, there's insufficient time to process the other 2 fireballs if they collided with something. They get skipped on the current frame and picked up on the next one. It's an OK compromise on a fireball that did collide with something, but if it didn't then the fireballs would end up traveling at variable speeds.

Hey...does this mean there's going to be a "brick walls" option now?

 

The texture pattern is hardcoded so that the AND takes 2 cycles. There's a chance I could make a 3 cycle AND work - but there's no RAM available to hold the texture patterns.

Nope - I'd have to check all 4 corners for each fireball on every frame they're in a castle row.

 

Perhaps code is too tight for this, but I was thinking of something like the following:

; Lower-left castle check--use similar code for other castles
; For simlicity, this code assumes bricks are4 coordinate-system pixels high.  I'm not sure how you have them stored in practice.
 lda ycoord
 lsr
 lsr
 tax
 rol
 and ycoord
 lsr  ; Carry set if two LSB's were '11'
 ldy ycoord
 bcs y_strad
y_nostrad:
 lda castlePF0,x
 and PF0mask,y
 bne hit
 lda caslePF1,x
 and PF1mask,y
 beq no_hit
 bne hit
y_strad:
 lda caslePF0,x
 ora castlePF0+1,x
 and PF0mask,y
 bne hit
 lda caslePF1,x
 ora caslePF1+1,x
 and PF1mask,x
 beq no_hit
 bne hit

 

This code will check to see if the ball has hit any of the castle bricks. It won't show which corner was hit, but I would think you could afford a bit more CPU time for the cases where the ball actually hit something.

Thanks. I'll keep that in mind if I need to address it.

The texture is better now. The brick lines either line up with the edge of the gradients or are in the center of them so it works visually.