GLSL Built-in Functions Quick Reference

1. Type Notation and genType

GLSL built-in signatures use generic type names. A function listed as taking genType actually has overloads for all of the following:

When a function takes genType x, genType y, both arguments must have the same type and dimension — you cannot mix vec3 and float without an explicit cast.

2. Angle and Trigonometry Functions

All trig functions operate on radians. Input/output are genType.

3. Exponential and Logarithm Functions

4. Common Math Functions

// Common snippet: smooth hash noise
float hash(vec2 p) {
  return fract(sin(dot(p, vec2(127.1, 311.7))) * 43758.5453);
}

// Safe smoothstep guard
float ss = smoothstep(0.2, 0.8, uv.x); // always e0 < e1

// Select between two values with step (no branch):
vec3 col = mix(colorA, colorB, step(0.5, uv.x));

5. Geometric Functions

// Phong diffuse + specular in one pass
vec3 N = normalize(vNormal);
vec3 L = normalize(uLightPos - vWorldPos);
vec3 V = normalize(uCamPos - vWorldPos);
vec3 R = reflect(-L, N);

float diff = max(dot(N, L), 0.0);
float spec = pow(max(dot(R, V), 0.0), uShininess);
vec3 color = diff * uDiffuse + spec * uSpecular;

6. Matrix Functions

7. Texture Lookup Functions

GLSL ES 3.0 uses sampler types and texture() overloads. GLSL ES 1.0 (WebGL 1) used texture2D(), textureCube() etc.

// Data texture (ping-pong simulation state):
uniform sampler2D uState;
uniform vec2 uRes;

vec4 readCell(ivec2 coord) {
  coord = clamp(coord, ivec2(0), ivec2(uRes) - 1);
  return texelFetch(uState, coord, 0);  // exact integer texel, no filtering
}

// Neighbour sum (Conway / reaction-diffusion):
ivec2 px = ivec2(gl_FragCoord.xy);
vec4 c = readCell(px);
vec4 L = readCell(px + ivec2(-1, 0));
vec4 R = readCell(px + ivec2( 1, 0));
vec4 U = readCell(px + ivec2( 0, 1));
vec4 D = readCell(px + ivec2( 0,-1));

8. Fragment Derivative Functions

Fragment derivative functions are only available in the fragment shader. They compute the rate of change of a value across neighbouring fragment quads.

Anti-aliased Shapes with fwidth

// Anti-aliased circle (no MSAA needed):
float circle(vec2 uv, float r) {
  float d = length(uv) - r;
  float w = fwidth(d);          // 1 pixel in distance units
  return 1.0 - smoothstep(-w, w, d);
}

// Anti-aliased grid lines:
float grid(vec2 uv, float spacing) {
  vec2 g = abs(fract(uv / spacing - 0.5) - 0.5);
  vec2 w = fwidth(uv / spacing);
  vec2 line = smoothstep(vec2(0.0), w * 1.5, g);
  return 1.0 - min(line.x, line.y);
}