Effects Missed by a Simple Ray-Tracer

Wave-lenght-dependence of the path of (refracted) light rays.

Caustics: Focussed bundles of many light rays.

   
--> We could start tracing light-rays from the light sources (this is what nature does);
-- but big efficiency problem: most rays nerver create any result seen by camera.

For objects or scenes of uniform color, our fake "ambient light" will produce useless pictures that show no features.

 
--> Ambient light does not reach all surface points with eaqual easy; put deeper shadows into narrow nooks and crannies.
--> Assume diffuse light comes from an overcast sky of uniform brightness; determine how much of that sky can be seen from any surface point.

Realistic global illumination must include diffuse reflection of light on other surfaces; e.g., indirect illumination behind cube.

 
--> Find an efficient way to approximate such indirect global illumination.  This is expensive!

If diffusely reflecting surfaces are colored, the secondary global illumination will also be colored; causing color-bleeding.

 
--> Take the color of surfaces into account when calculating the above diffusely reflected light components.
   

Clouds! -- and steam and smoke and fog ... What is the right geometry here?

--> Do we really want to model a million water droplets or dust particles?  This would be very expensive.
--> We need to use more efficient volumetric models, filled with some kind of participating medium.

Sun rays through uniform fog or haze or smoke.

 
--> Model this with a uniform sequence of light emitters along the path of a light ray through the participating medium.

Clouds of steam or smoke.

 
--> Use a non-uniform density function.  Calculate absorption of light as well as re-emission from the participating medium.