#version 420 #extension GL_ARB_texture_gather : enable // shader 8c1e55fd967b0496 // 1/4 -> 1/16 bloom pyramid . Pixelated unless scaled but still needs blur for light bleed.. // To-do. Check if screen res is * samples stable const float resScale = ($height/$gameHeight); // = 3.0; const int sampleScale = 3; const float lightBloom = 0.95; highp float lineRand(vec2 co) { highp float a = 12.9898; highp float b = 78.233; highp float c = 43758.5453; highp float dt = dot(co.xy, vec2(a, b)); highp float sn = mod(dt, 3.14); return fract(sin(sn) * c); } uniform ivec4 uf_remappedPS[5]; layout(binding = 0) uniform sampler2D textureUnitPS0;// Tex0 addr 0xf4e76000 res 320x180x1 dim 1 tm: 4 format 0816 compSel: 0 1 2 5 mipView: 0x0 (num 0x1) sliceView: 0x0 (num 0x1) Sampler0 ClampX/Y/Z: 6 6 6 border: 0 layout(location = 0) in vec4 passParameterSem0; layout(location = 0) out vec4 passPixelColor0; uniform vec2 uf_fragCoordScale; // FabriceNeyret2, single pass gaussian by intermediate MIPmap level. https://www.shadertoy.com/view/ltScRG // I hereby pledge my loyalty to the FabriceNeyret2 fanclub, this is bloody beautiful! const int samples = 8 * sampleScale, //8 or 4 balances xy position LOD = 2, // gaussian done on MIPmap at scale LOD sLOD = 1 << LOD; // tile size = 2^LOD const float sigma = float(samples) * .25; float gaussian(vec2 i) { return exp(-.5* dot(i /= sigma, i)) / (6.28 * sigma*sigma); } vec4 blur(sampler2D sp, vec2 U, vec2 scale) { vec4 O = vec4(0); int s = samples / sLOD; for (int i = 0; i < s*s; i++) { vec2 d = vec2(i%s, i / s)*float(sLOD) - float(samples) / 2.; O += gaussian(d) * textureLod(sp, U + scale * d, float(LOD)); } return O / O.a; } 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 R123f = 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; //R0f.xy = R0f.xy - (lineRand(gl_FragCoord.xy)*0.015); //R0f.xy = vec2((passParameterSem0.x + passParameterSem0.z), (passParameterSem0.y + passParameterSem0.w)); // 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[1].x) / resScale); R0f.w = R0f.y + (intBitsToFloat(uf_remappedPS[1].y) / resScale); R5f.w = 1.0; PS0f = R5f.w; // 1 R2f.x = R0f.x + (intBitsToFloat(uf_remappedPS[2].x) / resScale); R2f.y = R0f.y + (intBitsToFloat(uf_remappedPS[2].y) / resScale); R1f.z = R0f.x + (intBitsToFloat(uf_remappedPS[3].x) / resScale); R1f.w = R0f.y + (intBitsToFloat(uf_remappedPS[3].y) / resScale); // 2 backupReg0f = R0f.x; backupReg1f = R0f.y; R0f.x = backupReg0f + (intBitsToFloat(uf_remappedPS[4].x) / resScale); R0f.y = backupReg1f + (intBitsToFloat(uf_remappedPS[4].y) / resScale); vec2 coord = passParameterSem0.xy*textureSize(textureUnitPS0, 0); // R0f.xy;// vec2(0.5, 0.5); vec2 ps = vec2(1.0) / textureSize(textureUnitPS0, 0); vec2 uv = coord * ps; R4f.xyz = blur(textureUnitPS0, uv, ps*0.25).xyz; R4f.xyz += (texture(textureUnitPS0, R1f.xy).xyz)/2; R3f.xyz = (texture(textureUnitPS0, R0f.zw).xyz); R2f.xyz = (texture(textureUnitPS0, R2f.xy).xyz);// prob? R1f.xyz = (texture(textureUnitPS0, R1f.zw).xyz); R0f.xyz = (texture(textureUnitPS0, R0f.xy).xyz); /* //org R4f.xyz = (texture(textureUnitPS0, R1f.xy).xyz); R3f.xyz = (texture(textureUnitPS0, R0f.zw).xyz); R2f.xyz = (texture(textureUnitPS0, R2f.xy).xyz);// prob? R1f.xyz = (texture(textureUnitPS0, R1f.zw).xyz); R0f.xyz = (texture(textureUnitPS0, R0f.xy).xyz); */ // 0 PV0f.x = R4f.y * intBitsToFloat(uf_remappedPS[0].z); PV0f.y = R4f.x * intBitsToFloat(uf_remappedPS[0].z); PV0f.z = R4f.z * intBitsToFloat(uf_remappedPS[0].z); PV0f.w = max(R3f.x, R4f.x); R127f.z = max(R3f.y, R4f.y); PS0f = R127f.z; // 1 R123f.x = (R3f.z * intBitsToFloat(uf_remappedPS[1].z) + PV0f.z); PV1f.x = R123f.x; PV1f.y = max(R3f.z, R4f.z); R123f.z = (R3f.y * intBitsToFloat(uf_remappedPS[1].z) + PV0f.x); PV1f.z = R123f.z; R123f.w = (R3f.x * intBitsToFloat(uf_remappedPS[1].z) + PV0f.y); PV1f.w = R123f.w; R127f.y = max(PV0f.w, R2f.x); PS1f = R127f.y; // 2 R127f.x = max(R127f.z, R2f.y); R123f.y = (R2f.y * intBitsToFloat(uf_remappedPS[2].z) + PV1f.z); PV0f.y = R123f.y; R123f.z = (R2f.x * intBitsToFloat(uf_remappedPS[2].z) + PV1f.w); PV0f.z = R123f.z; R123f.w = (R2f.z * intBitsToFloat(uf_remappedPS[2].z) + PV1f.x); PV0f.w = R123f.w; R127f.w = max(PV1f.y, R2f.z); PS0f = R127f.w; // 3 backupReg0f = R0f.y; R123f.x = (R1f.x * intBitsToFloat(uf_remappedPS[3].z) + PV0f.z); PV1f.x = R123f.x; PV1f.y = max(R0f.x, R1f.x); R123f.z = (R1f.y * intBitsToFloat(uf_remappedPS[3].z) + PV0f.y); PV1f.z = R123f.z; R126f.w = (R1f.z * intBitsToFloat(uf_remappedPS[3].z) + PV0f.w); PS1f = max(backupReg0f, R1f.y); // 4 backupReg0f = R0f.x; backupReg1f = R0f.y; PV0f.x = max(R127f.y, PV1f.y); PV0f.y = max(R0f.z, R1f.z); R127f.z = (backupReg0f * intBitsToFloat(uf_remappedPS[4].z) + PV1f.x); PV0f.z = R127f.z; PV0f.w = max(R127f.x, PS1f); R127f.x = (backupReg1f * intBitsToFloat(uf_remappedPS[4].z) + PV1f.z); PS0f = R127f.x; // 5 backupReg0f = R0f.z; PV1f.x = PV0f.x + -(PV0f.z); PV1f.y = PV0f.w + -(PS0f); PV1f.z = max(R127f.w, PV0f.y); R127f.w = (backupReg0f * intBitsToFloat(uf_remappedPS[4].z) + R126f.w); PV1f.w = R127f.w; // 6 PV0f.x = PV1f.z + -(PV1f.w); R5f.y = (PV1f.y * 0.275 + R127f.x)*lightBloom;//degree of bloom R5f.x = (PV1f.x * 0.275 + R127f.z)*lightBloom; PS0f = R5f.x; // 7 R5f.z = (PV0f.x * 0.275 + R127f.w)*lightBloom; // export passPixelColor0 = vec4(R5f.x, R5f.y, R5f.z, R5f.w); }