"Physic"lly accurate
No its just more accurate in terms of how it uses color, its still no more physically accurate then you make use of. You can do this in mental ray if you will but then your the guy writing the shaders period.
Problem with using spectral rendering is that while the engine may calculate with spectra, normal modellers sure as hell are not able to use this info. See not only do you need more calculation time, but you'd need to be a very well tuned scientist to be able to model anything sane in this environment. See your shaders model something, but every time you have a shader youd need to pass a spectral color as opposed to a normal color. This in turn means taht you couldnt use any textures crated in Photoshop, you couldn't use cameras to capture your textures etc. Youd need specialized equipment built for you from scratch.
Second, to be physically accurate is a bit of a lie. Since we humans do NOT know how the physics of surfaces actually work, just how they approximately work. So if the physical accuracy of something like vray and menttalray would be on a scale form 1 -100. 10 Maxwell would be 10,1 at best. Even if you turn all the bells and whistles on the real limit to physical accuracy is the speed. Theres no magic in path tracing things its just very very slow. So we use tricks to do things at a sane speed to get good enough. Off course this brings us to the third problem, real world is not raytraced, but rather photons move from light sources towards the eye/camera. So even with sane speeds in maxwell its STILL a trick. And finally we do not know how the human auto exposure works, so any
So there is NO such thing as physically accurate, we have no freaking idea of how it really works, just better tuned versus little worse modeled. But its all up to the users skills mostly, so if your skill is 1 than the result on the 1-100 scale is STILL 1. Given that the rest of the way is so much work for you its a bit pointless. And even more its totally pointless to agrue about this if you dont write the shader code yourself, you lost 90% of any accuracy about your special need just there. There just needs to be some special justification to go there.
So how do you fake this?
Well simple you make a light that emits the spectra (it projects an image ramp) that tied to the light fog, emitting from within the prism. Because the maths involved is so easy you can with little effort calculate this spread. And its 100 tmes faster. But do you really need a lightfog for this effect? use a plane!? again 100 faster than the faked projection.
So at the end of the day you get more accuracy in terms of whaat you get to show whan you fake cartain things, so you can concentrate on whats really important.
PS: In any event when you have a justification for that kind of accuracy, that makes sense. Go to the local uinversity and ask for a grant of supercomputing time. I mean i have no special need and they still gave me 1 million processor hours of computation chips for developing a imaginary diesel engine.
