mirror of
https://github.com/cemu-project/cemu_graphic_packs.git
synced 2024-11-23 10:09:18 +01:00
079da6f21e
TPHD QL tweaks aka, sunday gaming delayed Resolution * Sharper, unscaled blur fixed. See contrasty note * Resolutions 4k and above scale sub viewports 75-> 50% of the main res. 7680x4320 is now 30 fps stable on a 1070, SSAA looks sweet. * Added some off-aspect custom res with horizontal res x2. Good for SSAA fixing shimmering as it's mostly horizontal in TPHD Contrasty * Sharper, unscaled blur fixed. Note, contrasty must be pre-selected at start to override same shader in resolution folder * Minor tweaks to cold setting Fancy FX *Added option to disable blur, just keeping other tweaks active. Realized that half scaling sub-view port mostly replaces fancy fX :/ *Tweaked to somewhat scale, tries to emulate original look by default. Ie 4x res = 4x blur "qualitative look" (not the same as nr of passes) *Added "defocus" looks blurrier by scaling incorrectly, masked by more draw passes. This is similar to the old default Not happy about the results of moving blur from res folder, too many drawbacks as quality of scaling really needs to be per-setting. Will probably deprecate if things start to break again. Maybe global variables in the future?
136 lines
4.1 KiB
Plaintext
136 lines
4.1 KiB
Plaintext
#version 420
|
|
#extension GL_ARB_texture_gather : enable
|
|
#extension GL_ARB_separate_shader_objects : enable
|
|
// shader c14019840473ff86
|
|
//scale boxblur fx
|
|
|
|
layout(binding = 0) uniform sampler2D textureUnitPS0;// Tex0 addr 0xf4001000 res 1920x1080x1 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 0xf5187000 res 1920x1080x1 dim 1 tm: 4 format 0011 compSel: 0 4 4 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;
|
|
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){ if( a == 0.0 || b == 0.0 ) return 0.0; return a*b; }
|
|
void main()
|
|
{
|
|
vec4 R0f = vec4(0.0);
|
|
vec4 R1f = vec4(0.0);
|
|
vec4 R2f = vec4(0.0);
|
|
vec4 R3f = vec4(0.0);
|
|
vec4 R123f = 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;
|
|
float scaler;
|
|
R0f = passParameterSem0;
|
|
scaler = uf_fragCoordScale.x;
|
|
R0f.w = (textureGather(textureUnitPS1, R0f.xy).y);
|
|
// 0
|
|
R1f.x = R0f.x + intBitsToFloat(0xba088889);
|
|
R1f.y = R0f.y + intBitsToFloat(0xba72b9d6);
|
|
R2f.z = R0f.x;
|
|
R2f.y = R0f.y + intBitsToFloat(0xba72b9d6);
|
|
PS0f = R2f.y;
|
|
// 1
|
|
R3f.x = R0f.x + intBitsToFloat(0xba088889);
|
|
R3f.y = R0f.y;
|
|
R1f.xyzw = (textureGather(textureUnitPS1, R1f.xy).xyzw);
|
|
R2f.yz = (textureGather(textureUnitPS1, R2f.zy).yz);
|
|
R3f.xy = (textureGather(textureUnitPS1, R3f.xy).xy);
|
|
// 0
|
|
PV0f.x = R2f.z + R3f.x;
|
|
PV0f.y = R1f.w + R0f.w;
|
|
PV0f.z = R1f.z + R3f.y;
|
|
PV0f.w = R1f.x + R2f.y;
|
|
// 1
|
|
R123f.x = (R1f.y * 2.0 + -(PV0f.x));
|
|
PV1f.x = R123f.x;
|
|
R123f.y = (R1f.y * 2.0 + -(PV0f.y));
|
|
PV1f.y = R123f.y;
|
|
R123f.z = (R1f.y * 2.0 + -(PV0f.z));
|
|
PV1f.z = R123f.z;
|
|
R123f.w = (R1f.y * 2.0 + -(PV0f.w));
|
|
PV1f.w = R123f.w;
|
|
// 2
|
|
PV0f.x = max(PV1f.x, -(PV1f.x));
|
|
PV0f.y = max(PV1f.y, -(PV1f.y));
|
|
PV0f.z = max(PV1f.z, -(PV1f.z));
|
|
PV0f.w = max(PV1f.w, -(PV1f.w));
|
|
// 3
|
|
PV1f.x = PV0f.x + intBitsToFloat(0xb400d959);
|
|
PV1f.y = PV0f.y + intBitsToFloat(0xb400d959);
|
|
PV1f.z = PV0f.z + intBitsToFloat(0xb400d959);
|
|
PV1f.w = PV0f.w + intBitsToFloat(0xb400d959);
|
|
// 4
|
|
R123f.x = intBitsToFloat(((PV1f.w >= 0.0)?(floatBitsToInt(1.0)):(0)));
|
|
PV0f.x = R123f.x;
|
|
R123f.y = intBitsToFloat(((PV1f.z >= 0.0)?(floatBitsToInt(1.0)):(0)));
|
|
PV0f.y = R123f.y;
|
|
R123f.z = intBitsToFloat(((PV1f.y >= 0.0)?(floatBitsToInt(1.0)):(0)));
|
|
PV0f.z = R123f.z;
|
|
R123f.w = intBitsToFloat(((PV1f.x >= 0.0)?(floatBitsToInt(1.0)):(0)));
|
|
PV0f.w = R123f.w;
|
|
// 5
|
|
tempf.x = dot(vec4(PV0f.x,PV0f.y,PV0f.z,PV0f.w),vec4(0.25*scaler,0.25*scaler,0.25*scaler,0.25*scaler));
|
|
PV1f.x = tempf.x;
|
|
PV1f.y = tempf.x;
|
|
PV1f.z = tempf.x;
|
|
PV1f.w = tempf.x;
|
|
// 6
|
|
R127f.z = PV1f.x * intBitsToFloat(0x3f400000);
|
|
PV0f.z = R127f.z;
|
|
// 7
|
|
R1f.x = (PV0f.z * intBitsToFloat(0xba088889) + R0f.x);
|
|
R1f.y = R0f.y;
|
|
R2f.z = R0f.x;
|
|
R2f.y = (PV0f.z * intBitsToFloat(0xba72b9d6) + R0f.y);
|
|
PS1f = R2f.y;
|
|
// 8
|
|
R3f.x = (R127f.z * intBitsToFloat(0x3a088889) + R0f.x);
|
|
R3f.y = R0f.y;
|
|
R0f.w = (R127f.z * intBitsToFloat(0x3a72b9d6) + R0f.y);
|
|
R1f.xyzw = (texture(textureUnitPS0, R1f.xy).xyzw);
|
|
R2f.xyzw = (texture(textureUnitPS0, R2f.zy).xyzw);
|
|
R3f.xyzw = (texture(textureUnitPS0, R3f.xy).xyzw);
|
|
R0f.xyzw = (texture(textureUnitPS0, R0f.xw).xyzw);
|
|
// 0
|
|
PV0f.x = R1f.w + R2f.w;
|
|
PV0f.y = R1f.z + R2f.z;
|
|
PV0f.z = R1f.y + R2f.y;
|
|
PV0f.w = R1f.x + R2f.x;
|
|
// 1
|
|
PV1f.x = R3f.w + PV0f.x;
|
|
PV1f.y = R3f.z + PV0f.y;
|
|
PV1f.z = R3f.y + PV0f.z;
|
|
PV1f.w = R3f.x + PV0f.w;
|
|
// 2
|
|
backupReg0f = R0f.y;
|
|
backupReg1f = R0f.x;
|
|
PV0f.x = R0f.w + PV1f.x;
|
|
PV0f.y = R0f.z + PV1f.y;
|
|
PV0f.z = backupReg0f + PV1f.z;
|
|
PV0f.w = backupReg1f + PV1f.w;
|
|
// 3
|
|
R0f.x = PV0f.w * 0.25;
|
|
R0f.y = PV0f.z * 0.25;
|
|
R0f.z = PV0f.y * 0.25;
|
|
R0f.w = PV0f.x * 0.25;
|
|
// export
|
|
passPixelColor0 = vec4(R0f.x, R0f.y, R0f.z, R0f.w);
|
|
}
|