It seems the sampling techniques I know from probabilistic ray tracing are showing up in attempts get smooth (and a bit noisy) effects like shadows, screen space ambient occlusion, …

I think it is time for us to have another look into deterministic low discrepancy sampling (quasi Monte Carlo), because it goes more naturally with current hardware based rasterization.

Atmospheric effects, fog, clouds, lightshafts, volumetric light in games …

Thinking about:

• improve the billboarding and volume feel of NinianeWangs approach to clouds
• mainly visible effects: out-scattering and coarse in-scattering
• coarse volume-based solution and detail particles/billboards
• probabilistic samples using available data (like variance shadow maps, ssao)
• path through volume is a line in a shadowmap - use for integration
• apply existing techniques for underwater effects to general participating media
• consider that spectrum is not just rgb and scattering is wavelength dependent
• analytically integrable fog density, tensor decompostion, screen space meshes
• depth peeling for combination with transparent objects and other effects
• painterly, cartoonish or ornamental clouds and fog
• …

At this place is planned:

A blog on random thoughts and information concerning computer graphics, algorithms, light transport, simulation, game ideas, personal projects, artwork and related topics.

Manuel Kugelmann