Earlier this year, there was a GSoC student that was interested in coding a spectral rendering mode for Cycles (something that wouldn’t be too unlike Luxrender), but it has been the truth for a while that the spectral effects that a spectral mode could bring to Cycles is notorious for being extremely slow to render (hence making such a mode less attractive for both existing and new developers and has been the subject of many slow convergence concerns on the Lux forums).
However, a new paper that has come out may mean that if developer coded a spectral render mode for Cycles and implemented chromatic aberration and dispersion shading for use with it, the effects will render out a lot faster than the techniques used today. (one of the figure descriptions talked about their new method allowing a level of convergence with 50 samples that used to require 900 samples).
For those who don’t understand the technical stuff, you can just look at the incredible speed differences in which spectral effects converge between this new method and the method used right now in engines like Luxrender, this means that such effects are now well within the range of being doable for artists or even animators without using the tried and true faking techniques.
The people on the Lux forum said it sounded as they were finally able to figure out how to do Multiple Importance Sampling with spectral samples, so this could be a big breakthrough if the main reason of avoidance by some is the slow convergence rate.
What are your thoughts?