I am quite confused as to how this output is produced. So is there some point light which is hemispherical, and for each point on the surface, it is checked whether that point "sees" light. If so it's irradiance is L(pi). This doesn't make sense because the lighting is definitely not just white or black. So how is the shading produced? If the light is bouncing, then how do we account for that?
motoole2
@anonymous Let's see if I can clear this up.
Short answer: This image refers to ambient occlusions (discussed in the following slide), and represents the amount of light that each point on the object receives from its environment.
Slightly longer answer: You can think of this image being generated in the following way. The illumination is uniformly distributed across all directions (it is not a point light source), and the appearance of the object is the result of evaluating the integral on the previous slide for every point under uniform illumination. The reason why this object does not appear pure white is that the integral from the previous slide is missing the so-called visibility term. That is, in concave regions of the object, the appearance is darker because some incident light directions are being blocked by the object itself. In convex regions, the appearance is brighter because few/no light directions are being blocked.
(Btw, ambient occlusions ignores the effect of light bouncing around.)
anonymous
Oh so I think I understand now. The "hemisphere" is taken as a hemisphere around the point we are considering (not the light source) in the direction normal to the surface and we integrate over L x 1* where 1* is the characteristic function of whether or not that "direction" is occluded?
I am quite confused as to how this output is produced. So is there some point light which is hemispherical, and for each point on the surface, it is checked whether that point "sees" light. If so it's irradiance is L(pi). This doesn't make sense because the lighting is definitely not just white or black. So how is the shading produced? If the light is bouncing, then how do we account for that?
@anonymous Let's see if I can clear this up.
Short answer: This image refers to ambient occlusions (discussed in the following slide), and represents the amount of light that each point on the object receives from its environment.
Slightly longer answer: You can think of this image being generated in the following way. The illumination is uniformly distributed across all directions (it is not a point light source), and the appearance of the object is the result of evaluating the integral on the previous slide for every point under uniform illumination. The reason why this object does not appear pure white is that the integral from the previous slide is missing the so-called visibility term. That is, in concave regions of the object, the appearance is darker because some incident light directions are being blocked by the object itself. In convex regions, the appearance is brighter because few/no light directions are being blocked.
(Btw, ambient occlusions ignores the effect of light bouncing around.)
Oh so I think I understand now. The "hemisphere" is taken as a hemisphere around the point we are considering (not the light source) in the direction normal to the surface and we integrate over L x 1* where 1* is the characteristic function of whether or not that "direction" is occluded?
@anonymous You got it!