The problem with IES data is that the profiles describe precisely how much light is emitted in which direction, i.e. they have a "shape" built in, but no correlating geometry. Light profiles assume that the light is an exact point light, and guarantee only that the IES/Eulumdat computation is correct at a sufficient distance. Soft shadows are not achievable with light profiles alone. The profiles are generated by measuring emitted light at the grid points of a grid placed around the light. For any light direction, this allows determining the light energy by interpolating the nearest grid points.
Maxwell uses physical geometry as emitters, and would require some special adaption (if it's even possible) to work with an emitter that somehow didn't have any physical geometry. Engines like mental ray let you assign an IES profile to an area light source, to approximate things like soft shadows from such profiles, but at that point, you're just approximating/guessing again.
In the end, I'm not sure how much of the IES standard is compatible with Maxwell's architecture, and I'm not sure if the traditional implementations are actually as accurate as you might get in Maxwell just by enterting explicit data for an accurate piece of geometry. I've not worked with IES profiles in anything other than mental ray, and since you have to assign it to area light sources to get soft shadows, you've corrupted the "accuracy" enough that you might as well have not used them in the first place, imo. Whether you could simply extract the spectral emission profile from a light independent of its shape is unknown - at least, to me, at this time.
_Mike
yay! hehe
- By Andreas Hopf