mirror/openlara
1
0
Fork 0
mirror of https://github.com/XProger/OpenLara.git synced 2025-08-06 21:26:56 +02:00
Code Issues Releases Wiki Activity

#368 GBA add extra 5 bits (13 in total) of precision for Gouraud shading, to reduce "saw" effect

Browse Source
This commit is contained in:
XProger
2022-04-24 04:17:27 +03:00
parent fedd2eda0c
commit d5f6139678
4 changed files with 27 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/platform/gba/asm/rasterizeFT.s
View File

@@ -48,8 +48,6 @@ dtmp .req t
Ltmp .req N Ltmp .req N
Rtmp .req N Rtmp .req N
Rti .req indexB
.macro PUT_PIXELS .macro PUT_PIXELS
tex indexA, t tex indexA, t
lit indexA lit indexA
@@ -176,17 +174,17 @@ rasterizeFT_asm:
ldrb indexB, [ptr, #-1]! // read pal index from VRAM (byte) ldrb indexB, [ptr, #-1]! // read pal index from VRAM (byte)
orr indexB, indexA, lsl #8 orr indexB, indexA, lsl #8
strh indexB, [ptr], #2 strh indexB, [ptr], #2
add t, dtdx
subs width, #1 // width-- subs width, #1 // width--
beq .scanline_end // if (width == 0) beq .scanline_end // if (width == 0)
add t, dtdx
.align_right: .align_right:
tst width, #1 tst width, #1
beq .align_block_4px beq .align_block_4px
sub Rti, Rt, dtdx tex indexA, Rt
tex indexA, Rti
lit indexA lit indexA
ldrb indexB, [ptr, width] ldrb indexB, [ptr, width]

8
src/platform/gba/asm/rasterizeFTA.s
View File

@@ -48,8 +48,6 @@ dtmp .req t
Ltmp .req N Ltmp .req N
Rtmp .req N Rtmp .req N
Rti .req indexB
.macro PUT_PIXELS .macro PUT_PIXELS
tex indexA, t tex indexA, t
add t, dtdx, lsl #1 add t, dtdx, lsl #1
@@ -179,17 +177,17 @@ rasterizeFTA_asm:
orrne indexB, indexA, lsl #8 orrne indexB, indexA, lsl #8
strneh indexB, [ptr], #2 strneh indexB, [ptr], #2
addeq ptr, #1 addeq ptr, #1
add t, dtdx
subs width, #1 // width-- subs width, #1 // width--
beq .scanline_end // if (width == 0) beq .scanline_end // if (width == 0)
add t, dtdx
.align_right: .align_right:
tst width, #1 tst width, #1
beq .align_block_4px beq .align_block_4px
sub Rti, Rt, dtdx tex indexA, Rt
tex indexA, Rti
cmp indexA, #0 cmp indexA, #0
ldrneb indexA, [LMAP, indexA] ldrneb indexA, [LMAP, indexA]

41
src/platform/gba/asm/rasterizeGT.s
View File

@@ -55,17 +55,17 @@ dtmp .req L
Ltmp .req N Ltmp .req N
Rtmp .req N Rtmp .req N
Rti .req tmp
Rgi .req tmp
SP_TILE = 0 SP_TILE = 0
SP_SIZE = 4 SP_SIZE = 4
.macro PUT_PIXELS G_EXTRA = 5 // extra bits of precision for gouraud shading (8 + G_EXTRA)
bic LMAP, g, #255
.macro PUT_PIXELS
tex indexA, t tex indexA, t
lit indexA
mov LMAP, g, lsr #(8 + G_EXTRA)
ldrb indexA, [indexA, LMAP, lsl #8]
strb indexA, [ptr], #2 // writing a byte to GBA VRAM will write a half word for free strb indexA, [ptr], #2 // writing a byte to GBA VRAM will write a half word for free
add g, dgdx, lsl #1 add g, dgdx, lsl #1
@@ -104,7 +104,7 @@ rasterizeGT_asm:
beq .calc_left_start beq .calc_left_start
lsl Lx, Lxy, #16 // Lx = L->v.x << 16 lsl Lx, Lxy, #16 // Lx = L->v.x << 16
lsl Lg, #8 // Lg <<= 8 lsl Lg, #(8 + G_EXTRA) // Lg <<= 8 + G_EXTRA
cmp Lh, #1 // if (Lh <= 1) skip Ldx calc cmp Lh, #1 // if (Lh <= 1) skip Ldx calc
beq .calc_left_end beq .calc_left_end
@@ -116,9 +116,9 @@ rasterizeGT_asm:
mul Ldx, tmp // Ldx = tmp * (N->v.x - Lx) mul Ldx, tmp // Ldx = tmp * (N->v.x - Lx)
ldrb Ldg, [N, #VERTEX_G] ldrb Ldg, [N, #VERTEX_G]
sub Ldg, Lg, lsr #8 sub Ldg, Lg, lsr #(8 + G_EXTRA)
mul Ldg, tmp // Ldg = tmp * (N->v.g - Lg) mul Ldg, tmp // Ldg = tmp * (N->v.g - Lg)
asr Ldg, #8 // 8-bit for fractional part asr Ldg, #(8 - G_EXTRA) // (8 + G_EXTRA)-bit for fractional part
ldr Ldt, [N, #VERTEX_T] ldr Ldt, [N, #VERTEX_T]
sub Ldt, Lt // Ldt = N->v.t - Lt sub Ldt, Lt // Ldt = N->v.t - Lt
@@ -143,7 +143,7 @@ rasterizeGT_asm:
beq .calc_right_start beq .calc_right_start
lsl Rx, Rxy, #16 // Rx = R->v.x << 16 lsl Rx, Rxy, #16 // Rx = R->v.x << 16
lsl Rg, #8 // Rg <<= 8 lsl Rg, #(8 + G_EXTRA) // Rg <<= 8 + G_EXTRA
cmp Rh, #1 // if (Rh <= 1) skip Rdx calc cmp Rh, #1 // if (Rh <= 1) skip Rdx calc
beq .calc_right_end beq .calc_right_end
@@ -155,9 +155,9 @@ rasterizeGT_asm:
mul Rdx, tmp // Rdx = tmp * (N->v.x - Rx) mul Rdx, tmp // Rdx = tmp * (N->v.x - Rx)
ldrb Rdg, [N, #VERTEX_G] ldrb Rdg, [N, #VERTEX_G]
sub Rdg, Rg, lsr #8 sub Rdg, Rg, lsr #(8 + G_EXTRA)
mul Rdg, tmp // Rdg = tmp * (N->v.g - Rg) mul Rdg, tmp // Rdg = tmp * (N->v.g - Rg)
asr Rdg, #8 // 8-bit for fractional part asr Rdg, #(8 - G_EXTRA) // (8 + G_EXTRA)-bit for fractional part
ldr Rdt, [N, #VERTEX_T] ldr Rdt, [N, #VERTEX_T]
sub Rdt, Rt // Rdt = N->v.t - Rt sub Rdt, Rt // Rdt = N->v.t - Rt
@@ -165,8 +165,8 @@ rasterizeGT_asm:
fiq_off fiq_off
.calc_right_end: .calc_right_end:
orr Lg, #LMAP_ADDR orr Lg, #(LMAP_ADDR << G_EXTRA)
orr Rg, #LMAP_ADDR orr Rg, #(LMAP_ADDR << G_EXTRA)
cmp Rh, Lh // if (Rh < Lh) cmp Rh, Lh // if (Rh < Lh)
movlt h, Rh // h = Rh movlt h, Rh // h = Rh
@@ -203,9 +203,9 @@ rasterizeGT_asm:
tst ptr, #1 // if (ptr & 1) tst ptr, #1 // if (ptr & 1)
beq .align_right beq .align_right
bic LMAP, g, #255
tex indexA, t tex indexA, t
lit indexA mov LMAP, g, lsr #(8 + G_EXTRA)
ldrb indexA, [indexA, LMAP, lsl #8]
ldrb indexB, [ptr, #-1]! // read pal index from VRAM (byte) ldrb indexB, [ptr, #-1]! // read pal index from VRAM (byte)
orr indexB, indexA, lsl #8 orr indexB, indexA, lsl #8
@@ -221,12 +221,9 @@ rasterizeGT_asm:
tst width, #1 tst width, #1
beq .align_block_4px beq .align_block_4px
sub Rti, Rt, dtdx tex indexA, Rt
tex indexA, Rti mov LMAP, Rg, lsr #(8 + G_EXTRA)
ldrb indexA, [indexA, LMAP, lsl #8]
sub Rgi, Rg, dgdx
bic LMAP, Rgi, #255
lit indexA
ldrb indexB, [ptr, width] ldrb indexB, [ptr, width]
subs width, #1 // width-- subs width, #1 // width--

9
src/platform/gba/asm/rasterizeGTA.s
View File

@@ -55,9 +55,6 @@ dtmp .req L
Ltmp .req N Ltmp .req N
Rtmp .req N Rtmp .req N
Rti .req tmp
Rgi .req tmp
SP_TILE = 0 SP_TILE = 0
SP_SIZE = 4 SP_SIZE = 4
@@ -229,15 +226,13 @@ rasterizeGTA_asm:
tst width, #1 tst width, #1
beq .align_block_4px beq .align_block_4px
sub Rti, Rt, dtdx tex indexA, Rt
tex indexA, Rti
cmp indexA, #0 cmp indexA, #0
subeq width, #1 subeq width, #1
beq .skip_right beq .skip_right
sub Rgi, Rg, dgdx, asr #1 bic LMAP, Rg, #255
bic LMAP, Rgi, #255
lit indexA lit indexA
ldrb indexB, [ptr, width] ldrb indexB, [ptr, width]