V-Flip Blit question

[Brick Nash]

Hi all,

I've been trying to flip a Bob vertically, and I'm almost there, but as you can see from the picture the colours are wrong.



Below are the tables I use to Blit an image to the screen. Each line is a horizontal slice of W-nn*H-16, all of which are stacked to make the image.

• The first LongWord is the address of the slice on the source Bitmap.
  
• The next two Words are H-Flip offsets for the slice (R & L)
  
• The next two Words are Source and Destination Modulos
  
• The final Word is the Blitsize.

DMO (Dest Modulo) is just the width of the Bitmap screen in bytes, and the width of the slice is subtracted. DMO changes when the screen width changes, so Bob tables don't need to be re-calculated (for when I was experimenting with screen sizes).

As you can see, I'm subtracting a negative slice width from DMO, but it's not quite there.

Any suggestions?

Many thanks!

FLIPPED IMAGE

	if 1=1
GUY_STANCE:			
	dc.w	$0000,$0000,VOS0+2,VOS0+2,$0022,-06-DMO,$1003
	dc.w	$0000,$0004,VOS1+4,VOS1+2,$0020,-08-DMO,$1004
	dc.w	$0000,$000A,VOS2+4,VOS2+4,$001E,-10-DMO,$1005
	dc.w	$0000,$0012,VOS3+4,VOS3+2,$0020,-08-DMO,$1004
	dc.w	$0000,$0018,VOS4+4,VOS4+2,$0020,-08-DMO,$1004
	dc.w	$0000,$001E,VOS5+4,VOS5+2,$0020,-08-DMO,$1004
	;---------------------------------------------------	
;	dc.w	$0000,$000C,$0000,$0000,$0024,DMO-04,$1002		; XY ORIGIN	
	;---------------------------------------------------
	dc.w	-1												; -1 Terminates count
	endif
	
UNFLIPPED IMAGE

	if 1=0
GUY_STANCE:			
	dc.w	$0000,$0000,VOS6+2,VOS6+2,$0022,DMO-06,$1003
	dc.w	$0000,$0004,VOS5+4,VOS5+2,$0020,DMO-08,$1004
	dc.w	$0000,$000A,VOS4+4,VOS4+4,$001E,DMO-10,$1005
	dc.w	$0000,$0012,VOS3+4,VOS3+2,$0020,DMO-08,$1004
	dc.w	$0000,$0018,VOS2+4,VOS2+2,$0020,DMO-08,$1004
	dc.w	$0000,$001E,VOS1+4,VOS1+2,$0020,DMO-08,$1004
	;---------------------------------------------------	
;	dc.w	$0000,$000C,$0000,$0000,$0024,DMO-04,$1002		; XY ORIGIN	
	;---------------------------------------------------
	dc.w	-1												; -1 Terminates count
	endif

[Samurai_Crow]

If you're trying to do interleaved blits with reverse modulo, the bitplane order will be reversed. Try non-interleaved blitting.

[Brick Nash]

Originally Posted by Samurai_Crow

If you're trying to do interleaved blits with reverse modulo, the bitplane order will be reversed. Try non-interleaved blitting.

Ok thanks, I'll give that a try.

[LeCaravage]

Interesting problem. I didn't turn things over in my head, so I could be wrong, but I see no reason why the blit does not work. Did you point $dff054 to the last interleaved bitplan ? And did you put negative modulos for source & destination ?
Do you blit each line or the whole bob in one pass ?

[a/b]

If you single-blit interleaved:
a0b0c0d0 a1b1c1d1 a2b2c2d2 => d2c2b2a2 d1c1b1a1 d0c0b0a0 (wanted: a2b2c2d2...) => lines+planes are flipped
If you single-blit non-interleaved:
a0a1a2 b0b1b2 c0c1c2 d0d1d2 => d2d1d0 c2c1c0 b2b1b0 a2a1a0 (wanted: a2a1a0...) => lines+planes are flipped
If you only want the lines flipped you have to do multiple blits, each plane separately, because single-blit modulo can't handle it.

[buzzybee]

Exactly, the sequence of bitplanes does not work anymore when blitting interleaved bitmaps with negative instead of positive modulo. I did a couple of tests some time ago, but only got tile flipping work on the x- and y-axis with CPU assistance.

One possible workaround is to adjust the palette the moment you flip a blitted object. But this is just an edge case that requires the object to use some color registers exclusively.

[Brick Nash]

Thanks for the advice everyone!

Looks like it's a bust with interleaved bitmaps, but I think I have enough memory to store the flipped frames I need, so all good.

I'm happy that I learned how it worked and what the limitations are anyway.