Question Does anyone know how to create an AABB for and SDF (Signed Distance field - procedural)

Hi everyone, I am using my own raymarcher, done in compute, with procedurally generated distance fields and was wondering if there was an easy way to create an AABB for groups of SDF's or as individual components. being that I was hoping to also include dynamic as SDF's are so robust.

so as a simple example if I use:

Code (CSharp):

1. //**********************************************************
2. // SPHERE
3. //**********************************************************
4. float iSphere(float3 p, float r)
5. {
6.     //------------------------------------------------------
7.     return length(p) - r;
8.     //------------------------------------------------------
9. }

obviously can become more complex once adding unions of shapes, but how would I create a bounding box, as it really doesn't exist except when it is being tested against with a ray march. doesn't need to be created in compute, so a c# example is fine. Can think of some ways to do it, but I think I am overthinking and not happy with my complex thoughts about it.

thanks in advance!

 

You can think of an AABB as an intersection with your map.
Max(AABB(),Map());
In the above the AABB and Map will be evaluated (no performance improvement).
Since you know that the AABB is going to be the more efficient test you can early out from it.
float aabb = AABB();
If (aabb < 0.01)
return Map();
else
return aabb;

A good process for working through the problem is start off with a blank AABB function that simply returns the same thing as your map.
Then you can start to think of your AABB function as a simplified version of your map function.
Now for every shape you added to your ABBB function determine the maximum point and minimum point (does not have to be a perfectly tight fit you can start with liberally large spheres). The algorithm for AABB building is simple you calculate the maximum and minimum points of your collection. In the case of a sphere you can do center + (radius, radius, radius) for the max and center - (radius, radius, radius).
Once you have the simple beginnings it becomes easier to optimize and explore ideas that are better catered for your implementation.

 

Ah yes, brilliant, thank you so much for taking the time to give me a starting point. So much simpler than what I was originally thinking, glad I asked before diving in).