How much more expensive are calculations for more complex materials compared to ideal specular surface for example?
Kent
I think they are much more expensive! I found a paper by Cornell University researchers entitled Discrete Stochastic Microfacet Models (http://www.cs.cornell.edu/projects/stochastic-sg14/stochastic.pdf), in which they explore the rendering of glittery surfaces (and I think that might be the paper that this picture was taken from as you can see from the first page of the paper :)). Glittery surfaces are complicated as they exhibit shifting random patterns of glints as the surface/viewer move relative to each other. Traditional techniques have utilized very high-resolution normal maps, however according to the authors these can create sever aliasing problems under "narrow-angle illumination". As a result, they propose a stochastic model for the effects of glitter that behave correctly at different distances and in motion. The model is based on micro facet theory. At the end of the paper they propose a multi scale procedural BRDF that can be implemented in standard rendering systems, which I think is pretty cool! They also note that in the limit their BRDF converges back to a smooth surface, which makes it very practical.
How much more expensive are calculations for more complex materials compared to ideal specular surface for example?
I think they are much more expensive! I found a paper by Cornell University researchers entitled Discrete Stochastic Microfacet Models (http://www.cs.cornell.edu/projects/stochastic-sg14/stochastic.pdf), in which they explore the rendering of glittery surfaces (and I think that might be the paper that this picture was taken from as you can see from the first page of the paper :)). Glittery surfaces are complicated as they exhibit shifting random patterns of glints as the surface/viewer move relative to each other. Traditional techniques have utilized very high-resolution normal maps, however according to the authors these can create sever aliasing problems under "narrow-angle illumination". As a result, they propose a stochastic model for the effects of glitter that behave correctly at different distances and in motion. The model is based on micro facet theory. At the end of the paper they propose a multi scale procedural BRDF that can be implemented in standard rendering systems, which I think is pretty cool! They also note that in the limit their BRDF converges back to a smooth surface, which makes it very practical.