[XCX] Aligment PS -> VS

[Update] Move alignment from PS to VS for more consistent behaviour
[Remove] Remove magic alignment values no longer needed in PS
This commit is contained in:
getdls 2017-11-22 00:04:44 +01:00
parent b3465f1025
commit c1d361fa4d
4 changed files with 62 additions and 290 deletions

View File

@ -54,8 +54,6 @@ void main()
vec3 cubeMapSTM; vec3 cubeMapSTM;
int cubeMapFaceId; int cubeMapFaceId;
R0f = passParameterSem0; R0f = passParameterSem0;
vec2 scaleFactor = vec2(repositionBloom, repositionBloom); //Added
R0f.xy = R0f.xy * scaleFactor;// re-align bloom.. hack until issue source is found.
// 0 // 0
R1f.x = R0f.x + (intBitsToFloat(uf_remappedPS[0].x) / resScale); R1f.x = R0f.x + (intBitsToFloat(uf_remappedPS[0].x) / resScale);

View File

@ -1,72 +0,0 @@
#version 420
#extension GL_ARB_texture_gather : enable
// shader 91f557914ed24b44
// re-alignt off axis bloom
//To-do Re-do whole chain with centered viewport values.
uniform ivec4 uf_remappedPS[1];
layout(binding = 0) uniform sampler2D textureUnitPS0;// Tex0 addr 0xf4386000 res 1280x720x1 dim 1 tm: 4 format 001a compSel: 0 1 2 3 mipView: 0x0 (num 0x1) sliceView: 0x0 (num 0x1) Sampler0 ClampX/Y/Z: 2 2 2 border: 0
layout(binding = 1) uniform sampler2D textureUnitPS1;// Tex1 addr 0xf5376000 res 640x360x1 dim 1 tm: 4 format 0816 compSel: 0 1 2 5 mipView: 0x0 (num 0x1) sliceView: 0x0 (num 0x1) Sampler1 ClampX/Y/Z: 2 2 2 border: 0
layout(location = 0) in vec4 passParameterSem0;
layout(location = 1) in vec4 passParameterSem1;
layout(location = 2) in vec4 passParameterSem2;
layout(location = 0) out vec4 passPixelColor0;
uniform vec2 uf_fragCoordScale;
const float repositionBloom = 1.00225;
int clampFI32(int v)
{
if( v == 0x7FFFFFFF )
return floatBitsToInt(1.0);
else if( v == 0xFFFFFFFF )
return floatBitsToInt(0.0);
return floatBitsToInt(clamp(intBitsToFloat(v), 0.0, 1.0));
}
float mul_nonIEEE(float a, float b){ return min(a*b,min(abs(a)*3.40282347E+38F,abs(b)*3.40282347E+38F)); }
void main()
{
vec4 R0f = vec4(0.0);
vec4 R1f = vec4(0.0);
vec4 R2f = vec4(0.0);
float backupReg0f, backupReg1f, backupReg2f, backupReg3f, backupReg4f;
vec4 PV0f = vec4(0.0), PV1f = vec4(0.0);
float PS0f = 0.0, PS1f = 0.0;
vec4 tempf = vec4(0.0);
float tempResultf;
int tempResulti;
ivec4 ARi = ivec4(0);
bool predResult = true;
vec3 cubeMapSTM;
int cubeMapFaceId;
R0f = passParameterSem0;
R1f = passParameterSem1;
R2f = passParameterSem2;
vec2 scaleFactor = vec2(repositionBloom, repositionBloom); //Added
R2f.xy = R2f.xy * scaleFactor;// re-align bloom.. hack until issue source is found.
R1f.xyz = (texture(textureUnitPS0, R1f.xy).xyz);
R2f.xyz = (texture(textureUnitPS1, R2f.xy).xyz);
// 0
tempf.x = dot(vec4(R1f.x,R1f.y,R1f.z,-0.0),vec4(intBitsToFloat(0x3e000000),intBitsToFloat(0x41ff0000),intBitsToFloat(0x45fe0100),0.0));
PV0f.x = tempf.x;
PV0f.y = tempf.x;
PV0f.z = tempf.x;
PV0f.w = tempf.x;
// 1
PV1f.y = -(R0f.x) + PV0f.x;
// 2
PV0f.x = max(PV1f.y, -(PV1f.y));
// 3
PV1f.w = R0f.y * PV0f.x;
PV1f.w = clamp(PV1f.w, 0.0, 1.0);
// 4
tempResultf = log2(PV1f.w);
if( isinf(tempResultf) == true ) tempResultf = -3.40282347E+38F;
PS0f = tempResultf;
// 5
PV1f.y = PS0f * intBitsToFloat(uf_remappedPS[0].w);
// 6
PS0f = exp2(PV1f.y);
// 7
R2f.w = PS0f * intBitsToFloat(uf_remappedPS[0].z); //sharpen
// export
passPixelColor0 = vec4(R2f.x, R2f.y, R2f.z, R2f.w);
}

View File

@ -0,0 +1,61 @@
#version 420
#extension GL_ARB_texture_gather : enable
#extension GL_ARB_shading_language_packing : enable
// shader 9dc2d340255dee89 //aligment vs
const float constvs = 8.0;
uniform ivec4 uf_remappedVS[1];
uniform vec2 uf_windowSpaceToClipSpaceTransform;
layout(location = 0) in uvec4 attrDataSem0;
layout(location = 1) in uvec4 attrDataSem1;
out gl_PerVertex
{
vec4 gl_Position;
float gl_PointSize;
};
layout(location = 0) out vec4 passParameterSem0;
int clampFI32(int v)
{
if( v == 0x7FFFFFFF )
return floatBitsToInt(1.0);
else if( v == 0xFFFFFFFF )
return floatBitsToInt(0.0);
return floatBitsToInt(clamp(intBitsToFloat(v), 0.0, 1.0));
}
float mul_nonIEEE(float a, float b){ return min(a*b,min(abs(a)*3.40282347E+38F,abs(b)*3.40282347E+38F)); }
void main()
{
vec4 R1f = vec4(0.0);
vec4 R2f = vec4(0.0);
uvec4 attrDecoder;
float backupReg0f, backupReg1f, backupReg2f, backupReg3f, backupReg4f;
vec4 PV0f = vec4(0.0), PV1f = vec4(0.0);
float PS0f = 0.0, PS1f = 0.0;
vec4 tempf = vec4(0.0);
float tempResultf;
int tempResulti;
ivec4 ARi = ivec4(0);
bool predResult = true;
vec3 cubeMapSTM;
int cubeMapFaceId;
attrDecoder.xyz = attrDataSem0.xyz;
attrDecoder.xyz = (attrDecoder.xyz>>24)|((attrDecoder.xyz>>8)&0xFF00)|((attrDecoder.xyz<<8)&0xFF0000)|((attrDecoder.xyz<<24));
attrDecoder.w = 0;
R1f = vec4(intBitsToFloat(int(attrDecoder.x)), intBitsToFloat(int(attrDecoder.y)), intBitsToFloat(int(attrDecoder.z)), intBitsToFloat(floatBitsToInt(1.0)));
attrDecoder.xy = attrDataSem1.xy;
attrDecoder.xy = (attrDecoder.xy>>24)|((attrDecoder.xy>>8)&0xFF00)|((attrDecoder.xy<<8)&0xFF0000)|((attrDecoder.xy<<24));
attrDecoder.z = 0;
attrDecoder.w = 0;
R2f = vec4(intBitsToFloat(int(attrDecoder.x)), intBitsToFloat(int(attrDecoder.y)), intBitsToFloat(floatBitsToInt(0.0)), intBitsToFloat(floatBitsToInt(1.0)));
// 0
backupReg0f = R2f.x;
backupReg1f = R2f.y;
R2f.x = (backupReg0f * intBitsToFloat(uf_remappedVS[0].x) + (intBitsToFloat(uf_remappedVS[0].z)/constvs));
R2f.y = (backupReg1f * intBitsToFloat(uf_remappedVS[0].y) + (intBitsToFloat(uf_remappedVS[0].w)/constvs));
// export
gl_Position = vec4(R1f.x, R1f.y, R1f.z, R1f.w);
// export
passParameterSem0 = vec4(R2f.x, R2f.y, R2f.z, R2f.z);
// 0
}

View File

@ -1,215 +0,0 @@
<?php
include 'Source/functions.php';
$fullWidth = $argv[1];
$fullHeight = $argv[2];
$scaleFactorX = always_decimal_format($fullWidth / 1280.0);
$scaleFactorY = always_decimal_format($fullHeight / 720.0);
?>
#version 420
#extension GL_ARB_texture_gather : enable
// shader b3fb199c73caa796
// open world and skell bloom.
// To-do - As long as brigntess fix is used, this will always be dodgy ..
const float resScale = <?=$scaleFactorX?>;
//const float resScale = 3.0;
//const float exposure = 0.9;
uniform ivec4 uf_remappedPS[4];
layout(binding = 0) uniform sampler2D textureUnitPS0;// Tex0 addr 0xf551a000 res 1280x720x1 dim 1 tm: 4 format 001a compSel: 0 1 2 3 mipView: 0x0 (num 0x1) sliceView: 0x0 (num 0x1) Sampler0 ClampX/Y/Z: 2 2 2 border: 0
layout(binding = 1) uniform sampler2D textureUnitPS1;// Tex1 addr 0xf470a000 res 1280x720x1 dim 1 tm: 4 format 0816 compSel: 0 1 2 5 mipView: 0x0 (num 0x1) sliceView: 0x0 (num 0x1) Sampler1 ClampX/Y/Z: 2 2 2 border: 0
layout(location = 0) in vec4 passParameterSem0;
layout(location = 0) out vec4 passPixelColor0;
uniform vec2 uf_fragCoordScale;
//const float repositionBloom = 1.00125;
//mattdesl MIT - https://github.com/Jam3/glsl-fast-gaussian-blur
vec4 blur9(sampler2D image, vec2 uv, vec2 resolution, vec2 direction) {
vec4 color = vec4(0.0);
vec2 off1 = vec2(1.3846153846) * direction;
vec2 off2 = vec2(3.2307692308) * direction;
color += texture2D(image, uv) * 0.2270270270;
color += texture2D(image, uv + (off1 / resolution)) * 0.3162162162;
color += texture2D(image, uv - (off1 / resolution)) * 0.3162162162;
color += texture2D(image, uv + (off2 / resolution)) * 0.0702702703;
color += texture2D(image, uv - (off2 / resolution)) * 0.0702702703;
return color;
}
int clampFI32(int v)
{
if( v == 0x7FFFFFFF )
return floatBitsToInt(1.0);
else if( v == 0xFFFFFFFF )
return floatBitsToInt(0.0);
return floatBitsToInt(clamp(intBitsToFloat(v), 0.0, 1.0));
}
float mul_nonIEEE(float a, float b){ return min(a*b,min(abs(a)*3.40282347E+38F,abs(b)*3.40282347E+38F)); }
void main()
{
vec4 R0f = vec4(0.0);
vec4 R1f = vec4(0.0);
vec4 R2f = vec4(0.0);
vec4 R3f = vec4(0.0);
vec4 R4f = vec4(0.0);
vec4 R5f = vec4(0.0);
vec4 R6f = vec4(0.0);
vec4 R7f = vec4(0.0);
vec4 R123f = vec4(0.0);
vec4 R124f = vec4(0.0);
vec4 R125f = vec4(0.0);
vec4 R126f = vec4(0.0);
vec4 R127f = vec4(0.0);
float backupReg0f, backupReg1f, backupReg2f, backupReg3f, backupReg4f;
vec4 PV0f = vec4(0.0), PV1f = vec4(0.0);
float PS0f = 0.0, PS1f = 0.0;
vec4 tempf = vec4(0.0);
float tempResultf;
int tempResulti;
ivec4 ARi = ivec4(0);
bool predResult = true;
vec3 cubeMapSTM;
int cubeMapFaceId;
R0f = passParameterSem0;
// 0
R1f.x = R0f.x + (intBitsToFloat(uf_remappedPS[0].x)/resScale);
R1f.y = R0f.y + (intBitsToFloat(uf_remappedPS[0].y) / resScale);
R0f.z = R0f.x + (intBitsToFloat(uf_remappedPS[0].z) / resScale);
R0f.w = R0f.y + (intBitsToFloat(uf_remappedPS[0].w) / resScale);
// 1
backupReg0f = R0f.x;
backupReg1f = R0f.y;
backupReg0f = R0f.x;
backupReg1f = R0f.y;
R0f.x = backupReg0f + (intBitsToFloat(uf_remappedPS[1].x) / resScale);
R0f.y = backupReg1f + (intBitsToFloat(uf_remappedPS[1].y) / resScale);
R1f.z = backupReg0f + (intBitsToFloat(uf_remappedPS[1].z) / resScale);
R1f.w = backupReg1f + (intBitsToFloat(uf_remappedPS[1].w) / resScale);
R2f.xyzw = (texture(textureUnitPS1, R1f.xy).xyzw);
R3f.xyzw = (texture(textureUnitPS1, R0f.zw).xyzw);
R4f.xyzw = (texture(textureUnitPS1, R0f.xy).xyzw);
R5f.xyzw = (texture(textureUnitPS1, R1f.zw).xyzw);
R6f.xyzw = (texture(textureUnitPS0, R1f.xy).xyzw);
R7f.xyzw = (texture(textureUnitPS0, R0f.zw).xyzw);
R0f.xyzw = (texture(textureUnitPS0, R0f.xy).xyzw);
R1f.xyzw = (texture(textureUnitPS0, R1f.zw).xyzw);
// 0
PV0f.x = R2f.z + R3f.z;
PV0f.y = R2f.y + R3f.y;
PV0f.z = R2f.x + R3f.x;
PV0f.w = R2f.w + R3f.w;
// 1
PV1f.x = PV0f.x + R4f.z;
PV1f.y = PV0f.y + R4f.y;
PV1f.z = PV0f.z + R4f.x;
PV1f.w = PV0f.w + R4f.w;
R127f.x = R6f.x + R7f.x;
PS1f = R127f.x;
// 2
PV0f.x = PV1f.x + R5f.z;
PV0f.y = PV1f.y + R5f.y;
PV0f.z = PV1f.z + R5f.x;
PV0f.w = PV1f.w + R5f.w;
R127f.w = R6f.y + R7f.y;
PS0f = R127f.w;
// 3
PV1f.x = PV0f.x * intBitsToFloat(uf_remappedPS[2].z);
PV1f.y = PV0f.y * intBitsToFloat(uf_remappedPS[2].z);
PV1f.z = PV0f.z * intBitsToFloat(uf_remappedPS[2].z);
PV1f.w = PV0f.w * intBitsToFloat(uf_remappedPS[2].z);
R127f.z = R6f.z + R7f.z;
PS1f = R127f.z;
// 4
PV0f.x = max(PV1f.x, 0.0);
PV0f.y = max(PV1f.y, 0.0);
PV0f.z = max(PV1f.z, 0.0);
PV0f.w = max(PV1f.w, 0.0);
R126f.y = R6f.w + R7f.w;
PS0f = R126f.y;
// 5
R6f.x = min(PV0f.x, 4.0);
PV1f.x = R6f.x;
R125f.y = min(PV0f.y, 4.0);
PV1f.y = R125f.y;
R126f.z = min(PV0f.z, 4.0);
PV1f.z = R126f.z;
R125f.w = min(PV0f.w, 4.0);
R124f.x = R127f.x + R0f.x;
PS1f = R124f.x;
// 6
PV0f.x = PV1f.z + -(intBitsToFloat(uf_remappedPS[2].x));
PV0f.y = PV1f.y + -(intBitsToFloat(uf_remappedPS[3].x));
PV0f.z = PV1f.z + -(intBitsToFloat(uf_remappedPS[3].x));
PV0f.w = PV1f.x + -(intBitsToFloat(uf_remappedPS[3].x));
R127f.y = PV1f.y + -(intBitsToFloat(uf_remappedPS[2].x));
PS0f = R127f.y;
// 7
R127f.x = R6f.x + -(intBitsToFloat(uf_remappedPS[2].x));
PV1f.y = max(PV0f.z, 0.0);
PV1f.z = max(PV0f.y, 0.0);
PV1f.w = max(PV0f.w, 0.0);
R126f.x = max(PV0f.x, 0.0);
PS1f = R126f.x;
// 8
backupReg0f = R127f.y;
R125f.x = dot(vec4(intBitsToFloat(uf_remappedPS[3].y),intBitsToFloat(uf_remappedPS[3].y),intBitsToFloat(uf_remappedPS[3].y),-0.0),vec4(PV1f.y,PV1f.z,PV1f.w,0.0));
PV0f.x = R125f.x;
PV0f.y = R125f.x;
PV0f.z = R125f.x;
PV0f.w = R125f.x;
R127f.y = max(backupReg0f, 0.0);
PS0f = R127f.y;
// 9
backupReg0f = R126f.y;
backupReg1f = R127f.x;
backupReg2f = R127f.w;
R127f.x = R127f.z + R0f.z;
R126f.y = backupReg0f + R0f.w;
PV1f.z = max(backupReg1f, 0.0);
R127f.w = backupReg2f + R0f.y;
R126f.w = R126f.z * PV0f.x;
PS1f = R126f.w;
// 10
tempf.x = dot(vec4(intBitsToFloat(uf_remappedPS[2].y),intBitsToFloat(uf_remappedPS[2].y),intBitsToFloat(uf_remappedPS[2].y),-0.0),vec4(R126f.x,R127f.y,PV1f.z,0.0));
PV0f.x = tempf.x;
PV0f.y = tempf.x;
PV0f.z = tempf.x;
PV0f.w = tempf.x;
R124f.w = tempf.x;
R127f.z = R125f.y * R125f.x;
PS0f = R127f.z;
// 11
PV1f.x = R125f.w * R125f.x;
PV1f.y = R6f.x * R125f.x;
R125f.z = R127f.w + R1f.y;
R127f.w = R127f.x + R1f.z;
R0f.w = R126f.y + R1f.w;
PS1f = R0f.w;
// 12
backupReg0f = R124f.x;
R124f.x = (R125f.y * R124f.w + R127f.z);
R125f.y = (R6f.x * R124f.w + PV1f.y);
PV0f.z = backupReg0f + R1f.x;
R123f.w = (R126f.z * R124f.w + R126f.w);
PV0f.w = R123f.w;
R127f.x = (R125f.w * R124f.w + PV1f.x);
PS0f = R127f.x;
// 13
R125f.x = R127f.w * intBitsToFloat(uf_remappedPS[3].z);
R126f.y = R125f.z * intBitsToFloat(uf_remappedPS[3].z);
PV1f.z = PV0f.z * intBitsToFloat(uf_remappedPS[3].z);
R127f.w = R0f.w * intBitsToFloat(uf_remappedPS[3].z);
PS1f = min(PV0f.w, intBitsToFloat(uf_remappedPS[2].w));
// 14
PV0f.x = min(R127f.x, intBitsToFloat(uf_remappedPS[2].w));
PV0f.y = min(R125f.y, intBitsToFloat(uf_remappedPS[2].w));
PV0f.z = min(R124f.x, intBitsToFloat(uf_remappedPS[2].w));
R0f.x = max(PV1f.z, PS1f);
PS0f = R0f.x;
// 15
R0f.y = max(R126f.y, PV0f.z);
R0f.z = max(R125f.x, PV0f.y);
R0f.w = max(R127f.w, PV0f.x);
// export
passPixelColor0 = vec4(R0f.x, R0f.y, R0f.z, R0f.w);
}