A Quick guide how to convert shader functions for Raymarching

Jon Baker, Graphics Programming is an incredible resources for shader functions, especially for signed distance fields.

This page explains how to convert these to use them with the [CustomSDF] operator.

// Original

  float de( vec3 p0 ){
    vec4 p = vec4(p0, 1.);
    for(int i = 0; i < 8; i++){
      p.xyz = mod(p.xyz-1.,2.)-1.;
      p*=1.4/dot(p.xyz,p.xyz);
    }
    return (length(p.xz/p.w)*0.25);
  }

What we need to do is to convert glsl into hlsl and replace the CustomSDF wrapper method...

The wrapper looks like this and is a build in part of the op:

// A wrapper function of the position p, the optional parameters Offset, A, B, and C
float dCustom(float3 p, float3 Offset, float A, float B, float C)
{

// This is the part you need to write into the shader parameter:
return 0;

} // The end of wrappers

So lets convert the code:

// Original

  float de( vec3 p0 ){                     // the original function needs to go away
    vec4 p = vec4(p0, 1.);                 // this adds a variable p, interfering with our parameter p
    for(int i = 0; i < 8; i++){            // stays
      p.xyz = mod(p.xyz-1.,2.)-1.;         // stays (but some taste magic numbers, that could be replaced with A B C)
      p*=1.4/dot(p.xyz,p.xyz);             // more magic numbers
    }
    return (length(p.xz/p.w)*0.25);        // more magic numbers
  }

So the first step would be to replace the ...

float4 p2 = float4(p, 1.);                 // We need to replace vec4 -> float4
for(int i = 0; i < 8; i++){            
    p2.xyz = mod(p2.xyz-1.,2.)-1.;         // use p2
    p*=1.4/dot(p2.xyz,p2.xyz);             // use p2
}
return (length(p2.xz/p2.w)*0.25);          // use p2

Now let's replace with some parameters. And also remove the glsl required float dots like 1..

Using parameters

float4 p2 = float4(p, 1);                 
for(int i = 0; i < 8; i++){            
    p2.xyz = mod(p2.xyz - 1, 2) - 1;        
    p*=1.4/dot(p2.xyz, p2.xyz);             
}
return (length(p2.xz/p2.w) * A);          // Let's use A so we can tweak it...

Keeping the defaults as comments might be a good idea...

float4 p2 = float4(p, 1.);
for(int i = 0; i < 8; i++){
    p2.xyz = mod(p2.xyz - 1, B) -1;   // B: 2
    p2*=C / dot(p2.xyz ,p2.xyz);      // C: 1.4
}
return ( length(p2.xz / p2.w) * A);   // A: 0.25

Adding a preset

Creating presets is even cooler, because it allows you to keep the original code and blend between paramters. (BTW. You may need to move your camera further away to see the fractal correctly)

Add credits and details

Even though Jon Bakers' shaders are CC0 non-commercial, it alwyas feels good to give credit:

//From https://jbaker.graphics/writings/DEC.html / unknown
float4 p2 = float4(p, 1.);
for(int i = 0; i < 8; i++){
    p2.xyz = mod(p2.xyz - 1, B) -1;   // B: 2
    p2*=C / dot(p2.xyz ,p2.xyz);      // C: 1.4
}
return ( length(p2.xz / p2.w) * A);   // A: 0.25