LYGIA Shader Library

noiseBlur (lygia/filter/noiseBlur)

Generic blur using a noise function inspired on https://www.shadertoy.com/view/XsVBDR

Dependencies:

Use:

noiseBlur(<SAMPLER_TYPE> texture, <vec2> st, <vec2> pixel, <float> radius)

Check it on Github




#ifndef NOISEBLUR_SAMPLES
#define NOISEBLUR_SAMPLES 4.0
#endif

#ifndef NOISEBLUR_TYPE
#define NOISEBLUR_TYPE vec4
#endif

#ifndef NOISEBLUR_SAMPLER_FNC
#define NOISEBLUR_SAMPLER_FNC(TEX, UV) SAMPLER_FNC(TEX, UV)
#endif

#ifndef NOISEBLUR_RANDOM23_FNC
#define NOISEBLUR_RANDOM23_FNC(UV) random2(UV)
#endif

#ifndef FNC_NOISEBLUR
#define FNC_NOISEBLUR

NOISEBLUR_TYPE noiseBlur(in SAMPLER_TYPE tex, in vec2 st, in vec2 pixel, float radius) {
    float blurRadius = radius;
    vec2 noiseOffset = st;

    #ifdef NOISEBLUR_SECS
    noiseOffset += 1337.0*fract(NOISEBLUR_SECS * 0.1);
    #endif

    NOISEBLUR_TYPE result = NOISEBLUR_TYPE(0.0);
    for (float i = 0.0; i < NOISEBLUR_SAMPLES; ++i) {

        #if defined(BLUENOISE_TEXTURE) && defined(BLUENOISE_TEXTURE_RESOLUTION)
        vec2 noiseRand = sampleNearest(BLUENOISE_TEXTURE, noiseOffset.xy, BLUENOISE_TEXTURE_RESOLUTION).xy;
        #else 
        vec2 noiseRand = NOISEBLUR_RANDOM23_FNC(vec3(noiseOffset.xy, i));
        #endif

        noiseOffset = noiseRand;

        vec2 r = noiseRand;
        r.x *= TAU;

        #if defined(NOISEBLUR_GAUSSIAN_K)
        // box-muller transform to get gaussian distributed sample points in the circle
        vec2 cr = vec2(sin(r.x),cos(r.x))*sqrt(-NOISEBLUR_GAUSSIAN_K * log(r.y));
        #else
        // uniform sample the circle
        vec2 cr = vec2(sin(r.x),cos(r.x))*sqrt(r.y);
        #endif

        NOISEBLUR_TYPE color = NOISEBLUR_SAMPLER_FNC(tex, st + cr * blurRadius * pixel );
        // average the samples as we get em
        // https://blog.demofox.org/2016/08/23/incremental-averaging/
        result = mix(result, color, 1.0 / (i+1.0));
    }
    return result;
}

NOISEBLUR_TYPE noiseBlur(SAMPLER_TYPE tex, vec2 st, vec2 pixel) {
    NOISEBLUR_TYPE rta = NOISEBLUR_TYPE(0.0);
    float total = 0.0;
    float offset = random(vec3(12.9898 + st.x, 78.233 + st.y, 151.7182));
    for (float t = -NOISEBLUR_SAMPLES; t <= NOISEBLUR_SAMPLES; t++) {
        float percent = (t / NOISEBLUR_SAMPLES) + offset - 0.5;
        float weight = 1.0 - abs(percent);
        NOISEBLUR_TYPE color = NOISEBLUR_SAMPLER_FNC(tex, st + pixel * percent);
        rta += color * weight;
        total += weight;
    }
    return rta / total;
}

#endif

Dependencies:

Use:

noiseBlur(<SAMPLER_TYPE> texture, <float2> st, <float2> pixel, <float> radius)

Check it on Github


#ifndef RANDOM_SCALE
#define RANDOM_SCALE float3(443.897, 441.423, .0973)
#endif
#ifndef RANDOM_SCALE_4
#define RANDOM_SCALE_4 float4(443.897, 441.423, .0973, 1.6334)
#endif



#ifndef NOISEBLUR_SAMPLES
#define NOISEBLUR_SAMPLES 4.0
#endif

#ifndef NOISEBLUR_TYPE
#define NOISEBLUR_TYPE float4
#endif

#ifndef NOISEBLUR_SAMPLER_FNC
#define NOISEBLUR_SAMPLER_FNC(TEX, UV) SAMPLER_FNC(TEX, UV)
#endif

#ifndef NOISEBLUR_RANDOM23_FNC
#define NOISEBLUR_RANDOM23_FNC(UV) random2(UV)
#endif

#ifndef FNC_NOISEBLUR
#define FNC_NOISEBLUR
NOISEBLUR_TYPE noiseBlur(in SAMPLER_TYPE tex, in float2 st, in float2 pixel, float radius) {
    float blurRadius = radius;
    float2 whiteNoiseUV = st;
    NOISEBLUR_TYPE result = float4(0.0, 0.0, 0.0, 0.0);
    for (float i = 0.0; i < NOISEBLUR_SAMPLES; ++i) {
        float2 whiteNoiseRand = NOISEBLUR_RANDOM23_FNC( float3(whiteNoiseUV.xy, i) );
        whiteNoiseUV = whiteNoiseRand;

        float2 r = whiteNoiseRand;
        r.x *= TAU;

        #if defined(NOISEBLUR_GAUSSIAN_K)
        // box-muller transform to get gaussian distributed sample points in the circle
        float2 cr = float2(sin(r.x),cos(r.x))*sqrt(-NOISEBLUR_GAUSSIAN_K * log(r.y));
        #else
        // uniform sample the circle
        float2 cr = float2(sin(r.x),cos(r.x))*sqrt(r.y);
        #endif

        NOISEBLUR_TYPE color = NOISEBLUR_SAMPLER_FNC(tex, st + cr * blurRadius * pixel );
        // average the samples as we get em
        // https://blog.demofox.org/2016/08/23/incremental-averaging/
        result = lerp(result, color, 1.0 / (i+1.0));
    }
    return result;
}

NOISEBLUR_TYPE noiseBlur(SAMPLER_TYPE tex, float2 st, float2 pixel) {
    NOISEBLUR_TYPE rta = float4(0.0, 0.0, 0.0, 0.0);
    float total = 0.0;
    float offset = random(float3(12.9898 + st.x, 78.233 + st.y, 151.7182));
    for (float t = -NOISEBLUR_SAMPLES; t <= NOISEBLUR_SAMPLES; t++) {
        float percent = (t / NOISEBLUR_SAMPLES) + offset - 0.5;
        float weight = 1.0 - abs(percent);
        NOISEBLUR_TYPE color = NOISEBLUR_SAMPLER_FNC(tex, st + pixel * percent);
        rta += color * weight;
        total += weight;
    }
    return rta / total;
}

#endif

Examples

License

LYGIA is dual-licensed under the Prosperity License and the Patron License for sponsors and contributors.

Sponsors and contributors are automatically added to the Patron License and they can ignore the any non-commercial rule of the Prosperity Licensed software (please take a look to the exception).

It's also possible to get a permanent comercial license hook to a single and specific version of LYGIA.

Get the latest news and releases

Sign up for the news letter bellow, joing the LYGIA's channel on Discord or follow the Github repository