How does it speed up computation if we assumed a spherical light source, since we don't typically have that? Does it help in the sense that we no longer have to consider how the model is occluding itself? We still have to compute lighting for all the sources in the scene, right?
So is the AO texture in addition to any lighting? So is the model basically being lit twice, once for hemispherical shadows and once for regular lighting? Does that make the shadows overdramatized?
A baked AO map makes certain assumptions about the lighting of a scene which may or may not always be true. Is it common practice to toggle AO maps or pre-bake multiple AO maps to account for variable lighting in a scene?
Yeah, I still don't totally understand the purpose of these maps. I can see that the hippo on the right looks better, but why doesn't the regular lighting in the scene already do that? Like, if the light is shining from above him, why ISN'T the inside of his mouth dark in the version without AO?
So, at least what I get from this slide is that there isn't actually a light in the scene, and we just pre-compute by assuming there is a light at infinity and factoring it into the color of the texture itself. I'm a bit confused as to what the distinction between hemispherical and spherical is, though. How would that make a difference? And why does it look like there is still some shading on the left picture?