Progressive stochastic ray tracing algorithms are increasingly used in
interactive applications such as design reviews and digital content
creation. However, convergence is still too slow to generate noise-free
images of interesting scenes at interactive rates. An approach that has
recently gained popularity is to use fast filtering techniques to reduce
the Monte Carlo noise. These methods can generate high-quality previews
or approximations of global illumination solutions fast enough to be
used in interactive applications.
In this talk, I will discuss recent developments in this direction,
including some of my own research. First, I will describe how classic
edge-aware filtering addresses the problem and demonstrate the strengths
and weaknesses of these approaches. Then I will discuss "radiance
filtering", a novel approach that alleviates some of the weaknesses and
provides visually more pleasing results in many situations.