Checkpoint for WIP scanline render fix for overlays+sprites and user-defined tiles

ORCACDefs/gte.h

@@ -155,7 +155,7 @@ extern pascal Word GTEClearOverlay() inline(0x23A0, tool_dispatcher);
 
 /* GTE Tile Constants */
 #define TILE_PRIORITY_BIT          0x4000                  /* Put tile on top of sprite */
-#define TILE_FRINGE_BIT            0x2000                  /* Unused */
+#define TILE_USER_BIT              0x2000                  /* User-defined tile */
 #define TILE_SOLID_BIT             0x1000                  /* Hint bit used in TWO_LAYER_MODE to optimize rendering */
 #define TILE_DYN_BIT               0x0800                  /* Is this a Dynamic Tile? */
 #define TILE_VFLIP_BIT             0x0400

macros/CORE.MACS.S

@@ -93,6 +93,16 @@ _PullReg2       mac
                 pla
                 <<<
 
+phxy            mac
+                phx
+                phy
+                <<<
+
+plyx            mac
+                ply
+                plx
+                <<<
+
 jne             mac
                 beq   *+5
                 jmp   ]1
@@ -103,6 +113,16 @@ jeq             mac
                 jmp   ]1
                 <<<
 
+jmi             mac
+                bpl   *+5
+                jmp   ]1
+                <<<
+
+jpl             mac
+                bmi   *+5
+                jmp   ]1
+                <<<
+
 jcc             mac
                 bcs   *+5
                 jmp   ]1
@@ -113,6 +133,12 @@ jcs             mac
                 jmp   ]1
                 <<<
 
+max             mac
+                cmp   ]1
+                bcs   mout
+                lda   ]1
+mout            <<<
+
 min             mac
                 cmp   ]1
                 bcc   mout

macros/GTE.Macs.s

@@ -135,6 +135,9 @@ _GTECompileSpriteStamp MAC
 _GTESetAddress       MAC
                      UserTool  $2E00+GTEToolNum
                      <<<
+_GTEUpdateOverlay    MAC
+                     UserTool  $2F00+GTEToolNum
+                     <<<
 
 ; EngineMode definitions
 ; Script definition
@@ -183,7 +186,7 @@ COPY_PIC_SCANLINE      equ   $0001      ; Copy in a way to support BG1 + RENDER_
 ; Tile constants
 ; TILE_RESERVED_BIT      equ   $8000
 TILE_PRIORITY_BIT      equ   $4000                  ; Put tile on top of sprite
-TILE_FRINGE_BIT        equ   $2000                  ; Unused
+TILE_USER_BIT          equ   $2000                  ; User-defined tile.  Execute registered callback.
 TILE_SOLID_BIT         equ   $1000                  ; Hint bit used in TWO_LAYER_MODE to optimize rendering
 TILE_DYN_BIT           equ   $0800                  ; Is this a Dynamic Tile?
 TILE_VFLIP_BIT         equ   $0400

src/Defs.s

@@ -179,12 +179,6 @@ RENDER_PER_SCANLINE    equ   $0010
 RENDER_WITH_SHADOWING  equ   $0020
 RENDER_SPRITES_SORTED  equ   $0040      ; Draw the sprites in y-sorted order.  Otherwise, use the index.
 
-; Overlay flags
-OVERLAY_MASKED         equ   $0000      ; Overlay has a mask, so the background must be draw first
-OVERLAY_SOLID          equ   $8000      ; Overlay covers the scan line and is fully opaque
-OVERLAY_ABOVE          equ   $0000      ; Overlay is drawn above scanline sprites
-OVERLAY_BELOW          equ   $4000      ; Overlay is drawn below scanline sprites
-
 ; DirtyBits definitions
 DIRTY_BIT_BG0_X        equ   $0001
 DIRTY_BIT_BG0_Y        equ   $0002
@@ -197,6 +191,7 @@ DIRTY_BIT_SPRITE_ARRAY equ   $0040
 ; GetAddress table IDs
 scanlineHorzOffset     equ   $0001        ; Table of 416 words, a double-array of scanline offset values. Values must be in range [0, 163]
 scanlineHorzOffset2    equ   $0002        ; Table of 416 words, a double-array of scanline offset values. Values must be in range [0, 163]
+tileStore              equ   $0003
 
 ; CopyPicToBG1 flags
 COPY_PIC_NORMAL        equ   $0000        ; Copy into BG1 buffer in "normal mode" treating the buffer as a 164x208 pixmap with stride of 256
@@ -239,6 +234,12 @@ SPRITE_8X8             equ   $0000                    ; 8 pixels wide x 8 pixels
 SPRITE_VFLIP           equ   $0400                    ; Flip the sprite vertically
 SPRITE_HFLIP           equ   $0200                    ; Flip the sprite horizontally
 
+; Overlay bits (aliases of SPRITE_ID bits)
+OVERLAY_MASKED         equ   $0000      ; Overlay has a mask, so the background must be draw first
+OVERLAY_SOLID          equ   $4000      ; Overlay covers the scan line and is fully opaque
+OVERLAY_ABOVE          equ   $0000      ; Overlay is drawn above scanline sprites
+OVERLAY_BELOW          equ   $2000      ; Overlay is drawn below scanline sprites
+
 ; Stamp storage parameters
 VBUFF_STRIDE_BYTES     equ {12*4}                        ; Each line has 4 slots of 16 pixels + 8 buffer pixels
 VBUFF_TILE_ROW_BYTES   equ {8*VBUFF_STRIDE_BYTES}        ; Each row is comprised of 8 lines
@@ -299,7 +300,9 @@ _ShadowListBottom EXT
 _DirectListCount  EXT
 _DirectListTop    EXT
 _DirectListBottom EXT
-
+ObjectListCount   EXT
+ObjectListHead    EXT
+ObjectList        EXT
 StartXMod164Tbl   EXT
 LastOffsetTbl     EXT
 BG1StartXMod164Tbl EXT

src/README.md

@@ -93,3 +93,43 @@ Possibilities
    lda  0000,y
    lda  0002,y
    ...
+
+= HOWTO: Custom tiles =
+
+   pea  #^MyTileRenderer         ; Define the callback function
+   pea  #MyTileRenderer
+   _GTESetCustomTileCallback
+
+   pea  0                        ; Put tiles with the TILE_USER_BIT set
+   pea  0
+   pea  TILE_ID+TILE_USER_BIT
+   _GTESetTile
+
+; On entry
+;
+; A = Tile Id. Use this to select what to draw
+; Y = Address of tile in the code field
+; X = Tile Store array offset
+; B = code field bank
+;
+; Use absolute addressing to fill in the code bank and address offsets $0000, $0003, $1000, $1003, ..., $7000, $7003. If
+; code needs to create snippets, it can be loaded from the tile store array by using the address returned from 
+; GTEGetAddress(tileStoreJmpAddr)
+;
+; User-defined tiles are always marked as damaged.
+
+MyTileRenderer
+   lda   #00A3                   ; Inserts LDA 00,s / PHA code to show a static background from Bank 0
+   sta:  $0000,y
+   sta:  $0003,y
+   sta:  $1000,y
+   sta:  $1003,y
+   ...
+   lda   #$4800
+   sta:  $0001,y
+   sta:  $0004,y
+   sta:  $1001,y
+   sta:  $1004,y
+   ...
+   rtl                           ; Must do a long return
+ 

src/SpriteRender.s

@@ -147,6 +147,10 @@ _DrawStampToScreen
              lda    _Sprites+IS_OFF_SCREEN,x        ; If the sprite is off-screen, don't draw it
              bne    _DSTSOut
 
+             lda    _Sprites+SPRITE_ID,x            ; If the sprite is hidden or an overlay, don't draw it
+             bit    #SPRITE_OVERLAY+SPRITE_HIDE
+             bne    _DSTSOut
+
              lda    _Sprites+SPRITE_CLIP_WIDTH,x    ; If the sprite is clipped to the playfield, don't draw it
              cmp    _Sprites+SPRITE_WIDTH,x
              bne    _DSTSOut

src/Tiles.s

@@ -374,6 +374,12 @@ _SetTile
 :changed         sta  oldTileId
                  lda  newTileId
                  sta  TileStore+TS_TILE_ID,x        ; Value is different, store it.
+
+; If the user bit is set, then skip most of the setup and just fill in the TileProcs with the user callback
+; target
+                 bit  #TILE_USER_BIT
+                 bne  :set_user_tile
+
                  jsr  _GetTileAddr
                  sta  TileStore+TS_TILE_ADDR,x      ; Committed to drawing this tile, so get the address of the tile in the tiledata bank for later
 
@@ -397,6 +403,26 @@ _SetTile
                  jsr  _SetNormalTileProcs
                  jmp  _PushDirtyTileX
 
+:set_user_tile
+                 lda  #UserTileDispatch
+                 stal K_TS_BASE_TILE_DISP,x
+                 lda  #UserTileDispatch
+                 stal K_TS_SPRITE_TILE_DISP,x
+                 lda  #UserTileDispatch
+                 stal K_TS_ONE_SPRITE,x
+                 jmp  _PushDirtyTileX
+
+; Trampoline / Dispatch table for user-defined tiles.  If the user calls the GTESetCustomTileCallback toolbox routine,
+; then this value is patched out.  Calling GTESetCustomTileCallback with NULL will disconnect the user's routine.
+UserTileDispatch
+                 ldal TileStore+TS_TILE_ID,x      ; Replace the tile data address (unset) with the tile id
+_UTDPatch        jsl  UserHook1                   ; Call the users code
+                 plb                              ; Restore the data bank
+                 rts
+
+; Stub to have a valid address for initialization / reset
+UserHook1        rtl
+
 ; X = Tile Store offset
 ; Y = Engine Mode Base Table address
 ; A = Table proc index
@@ -745,7 +771,7 @@ b_15_3     endbit 15;3;]4
 ; Store some tables in the K bank that will be used exclusively for jmp (abs,x) dispatch
 
 K_TS_BASE_TILE_DISP   ds TILE_STORE_SIZE      ; draw the tile without a sprite
-K_TS_COPY_TILE_DATA   ds TILE_STORE_SIZE      ; copy/merge the tile into temp storage
+;K_TS_COPY_TILE_DATA   ds TILE_STORE_SIZE      ; copy/merge the tile into temp storage
 K_TS_SPRITE_TILE_DISP ds TILE_STORE_SIZE      ; select the sprite routine for this tile
 K_TS_ONE_SPRITE       ds TILE_STORE_SIZE      ; specialized sprite routine when only one sprite covers the tile
-K_TS_APPLY_TILE_DATA  ds TILE_STORE_SIZE      ; move tile from temp storage into code field
+;K_TS_APPLY_TILE_DATA  ds TILE_STORE_SIZE      ; move tile from temp storage into code field

src/Tool.s

@@ -100,6 +100,7 @@ _CallTable
 
                 adrl  _TSCompileSpriteStamp-1
                 adrl  _TSSetAddress-1
+                adrl  _TSUpdateOverlay-1
 
 _CTEnd
 _GTEAddSprite        MAC
@@ -721,6 +722,7 @@ _TSSetOverlay
 
                 ldx     #0                              ; Always just use the first spot
                 lda     #SPRITE_OVERLAY
+;                lda     #SPRITE_OVERLAY+SPRITE_HIDE     ; Type 2 overlays re-use the HIDE bit
                 sta     Overlays+OVERLAY_ID,x
                 stz     Overlays+OVERLAY_FLAGS,x
 
@@ -739,7 +741,7 @@ _TSSetOverlay
                 lda     :proc+2,s
                 sta     Overlays+OVERLAY_PROC+2,x
 
-                ldx     #{MAX_SPRITES+0}*2              ; Adjust to call the generic routings
+                ldx     #{MAX_SPRITES+0}*2              ; Adjust to call the generic routine
                 jsr     _InsertSprite
  
                 _TSExit #0;#8
@@ -753,15 +755,25 @@ _TSUpdateOverlay
                 
                 ldx     #0
                 stz     Overlays+OVERLAY_FLAGS,x
+
                 lda     :top,s
                 sta     Overlays+OVERLAY_TOP
                 lda     :bottom,s
+                dec
                 sta     Overlays+OVERLAY_BOTTOM
+                sec
+                sbc     :top,s
+                inc
+                sta     Overlays+OVERLAY_HEIGHT,x
+                
                 lda     :proc,s
                 sta     Overlays+OVERLAY_PROC
                 lda     :proc+2,s
                 sta     Overlays+OVERLAY_PROC+2
 
+                ldx     #{MAX_SPRITES+0}*2              ; Adjust to call the generic routings
+                jsr     _UpdateSprite
+
                 _TSExit #0;#8
 
 ; ClearOverlay()

src/blitter/Blitter.s

@@ -13,6 +13,11 @@ _BltRange
 :exit_ptr       equ   tmp0
 :jmp_low_save   equ   tmp2
 
+                sty   :jmp_low_save  ; Steal some temp space and check for empty ranges
+                cpx   :jmp_low_save  ; This makes it simpler for some callers
+                bcc   *+3
+                rts
+
                 phb                  ; preserve the bank register
                 clc`
 

src/blitter/PEISlammer.s

@@ -13,19 +13,26 @@
 ; X = first line (inclusive), valid range of 0 to 199
 ; Y = last line  (exclusive), valid range >X up to 200
 _PEISlam
-                 cpx   #201
+                 cpx   #200
-                 bcc   *+4
+                 bcc   *+3
-                 brk   $A9
+                 rts
                  cpy   #201
-                 bcc   *+4
+                 bcc   *+3
-                 brk   $A8
+                 rts
+
+                 txa
+                 stal  :screen_width_1
+                 tya
+                 cmpl  :screen_width_1
+                 bcs   *+3
+                 rts
+
 
                  lda   ScreenWidth
                  dec
                  stal  :screen_width_1  ; save the width-1 outside of the direct page
 
                  lda   #:pei_end        ; patch the PEI entry address
-                 and   #$FFFE           ; should always be even, but....
                  sec
                  sbc   ScreenWidth
                  stal  :inner+1

src/blitter/Rotation.s

@@ -6,8 +6,13 @@
 ; 
 ; This is about as fast of a rotation as we can do.
 ;
-; When possible, off-screen locations are calculate to produce an address of $FFFE, so that the last two bytes
+; When possible, off-screen locations are calculated to produce an address of $FFFE, so that the last two bytes
 ; of the BG1 data buffer provides the "fill value".
+;
+; Having a fixed table of addresses is limiting due to the size an inability to control what happens at the
+; boundaries. Consider generating some pre-processed step + error parameters that can be used in a fast
+; DDA stepper to allow a more compact representation or different (scale, angle) pairs.  This could allow for
+; a full range of 256 rotation angles + multiple scalings.
 
 ANGLEBNK                   EXT
 _ApplyBG1XPosAngle

src/static/TileStore.s

@@ -407,6 +407,7 @@ OldOneSecVec      ENT
 Timers            ENT
                   ds        TIMER_REC_SIZE*MAX_TIMERS
 
+; Keep a count of the different overlays
 ;Overlays          ENT
 ;                  dw        0     ; count
 ;                  ds        10    ; only support one for now (flags, start_line, end_line, function call)
@@ -543,6 +544,14 @@ _DirectListTop    ENT
 _DirectListBottom ENT
                   ds {2*{MAX_ELEMENTS+1}}
 
+; List of filtered objects
+ObjectListCount    ENT
+                  ds  2
+ObjectListHead    ENT
+                  ds  2
+ObjectList        ENT
+                  ds  {10*{MAX_ELEMENTS+2}}   ; Extra space at the end for a sentinel marker
+
 
 ; Steps to the different sprite stamps
 _stamp_step      ENT

src/static/TileStoreDefs.s

@@ -31,6 +31,13 @@ SPRITE_REC_SIZE        equ 42
 MAX_OVERLAYS           equ 2
 MAX_ELEMENTS           equ {MAX_SPRITES+MAX_OVERLAYS}
 
+; Object list used in  renderer
+OL_SPRITE_ID           equ 0                     ; Usual parallel arrays
+OL_SPRITE_TOP          equ {2*{MAX_ELEMENTS+1}}
+OL_SPRITE_BOTTOM       equ {4*{MAX_ELEMENTS+1}}
+OL_NEXT                equ {6*{MAX_ELEMENTS+1}}
+OL_INDEX               equ {8*{MAX_ELEMENTS+1}}  ; Reference to the index in the _Sprites array
+
 ; Mark each sprite as ADDED, UPDATED, MOVED, REMOVED depending on the actions applied to it
 ; on this frame.  Quick note, the same Sprite ID cannot be removed and added in the same frame.
 ; A REMOVED sprite if removed from the sprite list during the Render call, so it's ID is not