; A 123-byte Apple II Lo-res Fake Palette Rotation Demo ; The Apple II has no Palette rotation hardware, so we fake it ; For Lovebyte 2021 ; by Vince `deater` Weaver (vince@deater.net) / dSr ; with some help from qkumba ; 151 -- original ; 137 -- optimize generation ; 136 -- align lookup table so we can index it easier ; 130 -- optimize indexing of lookup ; 126 -- run loops backaward ; 124 -- notice X already 0 before plot ; 131 -- use GBASCALC. much faster, but 7 bytes larger ; 129 -- run loop backwards ; 128 -- set color ourselves ; 127 -- overlap color lookup with sine table ; 119 -- forgot to comment out unused ; 121 -- make it use full screen (40x48) ; 149 -- add page flipping ; 144 -- optimize a bit ; 141 -- smc DRAW_PAGE ; 139 -- from qkumba, remove php/plp ; 138 -- from qkumba, remove SAVEX ; 133 -- run from zero page ; 132 -- make lookup 8*sin+7 ; 131 -- re-arrange sine table ; 128 -- call into PLOT for MASK seting ; urgh lovebyte wants 124 byte (counts header) ; 127 -- base YY<<16 by adding smc, not by shifting ; 125 -- realize that the top byte wraps so no need to and ; 124 -- re-arrange code to make an CLC unnecessary ; 123 -- qkumba noticed we can use the $FF offset directly in page flip ; zero page GBASL = $26 GBASH = $27 MASK = $2E COLOR = $30 ;CTEMP = $68 YY = $69 ; soft-switches FULLGR = $C052 PAGE1 = $C054 ; ROM routines PLOT1 = $F80E ;; PLOT at (GBASL),Y (need MASK to be $0f or $f0) GBASCALC= $F847 ;; take Y-coord/2 in A, put address in GBASL/H ( a trashed, C clear) SETCOL = $F864 ;; COLOR=A*17 SETGR = $FB40 .zeropage .globalzp colorlookup,plot_lookup_smc,draw_page_smc,frame_smc,sinetable ;================================ ; Clear screen and setup graphics ;================================ plasma: jsr SETGR ; set lo-res 40x40 mode bit FULLGR ; make it 40x48 ; color = ( 8.0 + 8*sin(x) + 8.0 + 8*sin(y) )/2 ; becomes ; color = ( 16 + (sintable[xx&0xf]) + (sintable[yy&0xf])) / 2; ; we only create a 16x16 texture, which we pattern across 40x48 screen ; I've tried re-optimizing this about 10 different ways ; and it never ends up shorter create_lookup: ldx #15 create_yloop: ldy #15 create_xloop: sec lda sinetable,X adc sinetable,Y ; 15+sin(x)+sin(y) lsr lookup_smc: sta lookup ; always starts at $d00 inc lookup_smc+1 dey bpl create_xloop dex bpl create_yloop ; X and Y both $FF create_lookup_done: forever_loop: cycle_colors: ; cycle colors ; instead of advancing entire frame, do slightly slower route ; instead now and just incrememnting the frame and doing the ; adjustment at plot time. ; increment frame inc frame_smc+1 ; set/flip pages ; we want to flip pages and then draw to the offscreen one flip_pages: ; ldy #0 ; iny ; y is $FF, make it 0 lda draw_page_smc+1 ; DRAW_PAGE bne done_page dey done_page: ; ldx PAGE1,Y ; set display page to PAGE1 or PAGE2 ldx $BF56,Y ; PAGE1 - $FF eor #$4 ; flip draw page between $400/$800 sta draw_page_smc+1 ; DRAW_PAGE ; plot current frame ; scan whole 40x48 screen and plot each point based on ; lookup table colors plot_frame: ldx #47 ; YY=47 (count backwards) plot_yloop: txa ; get YY into A pha ; save X for later lsr ; call actually wants Ycoord/2 php ; save C flag for mask handling ; ugh can't use PLOT trick as it always will draw something ; to PAGE1 even if we don't want to jsr GBASCALC ; point GBASL/H to address in (A is ycoord/2) ; after, A is GBASL, C is clear lda GBASH ; adjust to be PAGE1/PAGE2 ($400 or $800) draw_page_smc: adc #0 sta GBASH ; increment YY in top nibble of lookup for (yy<<16)+xx ; clc from above, C always 0 lda plot_lookup_smc+1 adc #$10 ; no need to mask as it will oflo and be ignored sta plot_lookup_smc+1 ;========== ldy #39 ; XX = 39 (countdown) ; sets MASK by calling into middle of PLOT routine ; by Y being 39 draw in a spot that gets over-written plp jsr $f806 plot_xloop: tya ; get XX & 0x0f and #$f tax plot_lookup_smc: lda lookup,X ; load lookup, (YY*16)+XX clc frame_smc: adc #$00 ; add in frame and #$f lsr ; we actually only have 8 colors tax lda colorlookup,X ; lookup color sta COLOR ; each nibble should be same jsr PLOT1 ; plot at GBASL,Y (x co-ord goes in Y) dey bpl plot_xloop pla ; restore YY tax dex bpl plot_yloop bmi forever_loop colorlookup: ; blue .byte $55,$22,$66,$77,$ff,$77,$55 ; ,$00 shared w sin table sinetable: ; this is actually (8*sin(x))+7 ; re-arranged so starts with $00 for colorlookup overlap .byte $00,$FF HACK: ; use the $0200 here for (HACK),Y addressing? ; in the end no way to get Y set properly .byte $00,$02,$04 .byte $07,$0A,$0C,$0E,$0F,$0E,$0C,$0A .byte $07,$04,$02 ; make lookup happen at page boundary .org $200 lookup: