From Pixar's paper
This slice of hair shell is based on FMM (Fast Marching Method) to obain A Level Set function in order to get a "imaginary" curve normal for hair shading similar to "volumetric methods for simulation and rendering of hair" from Pixar studio. but I really do not know whether we need this because PBR hair rendering do not need normal at all, this direction is Pixar-only. I have also implement the Marschner's model, viewed the dual scattering published on Siggraph 2008, but I am still confused with the real need of production.
This 2D slice is obtained based on the level set of FMM to solve the density function. It is equivalent to Pixar's volumetric methods for simulation and rendering of hair.Article. However, I am still confused about what the real production needs are and what technologies are needed to meet the results requirements. I also did the physical model of Marschner. I also saw the bidirectional scattering hair rendering at the Bonn University in Siggraph 08 this year, and I felt that the gap with production was too big.
It means shadow is required. Well, deep shadow is acceptable, but the bias value on the body is incorrect. RT, no, but the speed is unacceptable. Even adding colors is not easy to add, modeling is not easy to build, the dynamics part also needs character TD multi-study, in short, the road is long.