When we have directional lighting, we are also assuming a light source at infinity. But in this case, even though we again have a light source at infinity, are we saying that the light source is approximately an infinitely large sphere centered about the object we are interested in?
This requires us to assume a spherical light. In practice, particularly in gaming, how useful is this technique since most light sources don't seem to be spherical lights?
How can this information be efficiently stored and recomputed for a changing light source?
Does this mean this needs to get recomputed when the model moves?
When we look at a different position, the light should be changed in direction and needs more computation, is it?
Do we take reflections into account when generating ambient occlusion map?
Does this only take into account one light source? How would calculations change if there were multiple light sources, or if some material was reflective?
The value of this pre-computed map depends on where we put the assumed light source right? How do we decide which direction our lights are coming from?
What information do we bake into the pre-computed texture? How is this information useful when the lighting sources change?
What algorithm allows us to pre-compute the occlusion when making the texture?
How can the pre-computation be helpful and does the pre-computation directly affect the map?
What do we do when the light isn't spherical?
Can this be expressed as mipmap as well?
Is this technique only applicable for pre-computing on textures and how would this generalize or react to different light sources?
Can we also store color values as well when taking this into account? or is there some reason this couldn't be done well in practice?
What data structures allow us to efficiently calculate ambient occlusion?
The ambient occlusion map seems quite specific, only applying to spherical lighting. Is it actually applicable with other forms of lighting?
The model for ambient occlusion still doesn't make much sense yet, is it possible to go over this more? I understand intuitively how this might be computed, but is there a specific algorithm for it?
Does ambient occlusion only work with spherical lighting? I don't understand how with a different or changing light source how ambient occlusion could be efficiently calculated or transformed.