Can we get more information and or resources to study how this affect is generated? Was this same effect used in the Pixar movie Monsters Inc?
weinatalie
The rendering models shown in lecture appear to take in each strand of hair as an individual cylinder, but I feel like it would be very computationally taxing to have to model every single strand of hair or fur. I looked more into hair modeling and found that hair is usually modeled in one of three ways: using curves or splines to represent individual strands, layering several polygons to simulate layered strips of hair, or particle systems. If, to render hair, one was simply mapping a hair texture to a polygon, I wonder how you could achieve the extra dimension like the model in the image. Could you use a displacement map to convey the hair’s coarseness, or add in smaller details retroactively such as flyaway strands?
carolyn-wang
Pixar has an interesting rendering algorithm for stabling simulating hair. I remember reading about it when Brave came out. The curly hair model is based on an extensible elastic rod. They also parallelize movement of hairs to make the simulation more realistic. Here’s a link to the paper:
https://graphics.pixar.com/library/CurlyHairA/paper.pdf
aravmisra
@carolyn-wang that is SUPER COOL that you brought up pixar- its exactly what I thought of for modeling hair, since I recall hearing that hair was one of the most difficult things to animate in early pixar (hence why the first movie, Toy Story, featured toys).
A related article/interview talks about how the hair in Monsters University was so detailed that it required massive farming and rendering jobs! https://venturebeat.com/games/the-insiders-view-of-the-tech-behind-pixars-monsters-university-interview/
aidangarde
I assume most hair is not a sequence of strands with material data, but rather a shape with a hair texture. How can you render light reflecting of hair correctly without representing each strand? My guess would have a pattern computed across all the hairs, then apply the texture to represent each hair.
stang085
I can't image how much time it would take to calculate all the different hairs, since each of them give off such different looks, and depending on each part of the hair, it's shiny, and the light passes through it. It's also interesting how the light catches each individual hair strand, and it's hard to simplify it since you can tell that each strand should react differently to the light
muuncakez
For Disney, with Rapunzel, her hair was represented with dynamic wires. This allowed different laws of physics to give it this more free form, flowing look. I thought it was interesting that they seemingly then just applied a hair texture and BRDF to these wires to further convince the eye this is hair (its shiny like the hair appearance here, differs in shading, etc). Furthermore, her head of hair was mapped as 147 tubes instead of animating around 140,000 strands of hair. The tubes themselves had a bunch of "breaking points" or "joints" that the tube would bend at to give animators more freedom in what form her hair took. I presume this also lessens the computational tax to render her hair and creates a more convincing finish.
KevinXu02
One most simpliest way is to layer textures to create the illusion of fur. As furs are thin and transparent, it is always hard to render them correctly.
Can we get more information and or resources to study how this affect is generated? Was this same effect used in the Pixar movie Monsters Inc?
The rendering models shown in lecture appear to take in each strand of hair as an individual cylinder, but I feel like it would be very computationally taxing to have to model every single strand of hair or fur. I looked more into hair modeling and found that hair is usually modeled in one of three ways: using curves or splines to represent individual strands, layering several polygons to simulate layered strips of hair, or particle systems. If, to render hair, one was simply mapping a hair texture to a polygon, I wonder how you could achieve the extra dimension like the model in the image. Could you use a displacement map to convey the hair’s coarseness, or add in smaller details retroactively such as flyaway strands?
Pixar has an interesting rendering algorithm for stabling simulating hair. I remember reading about it when Brave came out. The curly hair model is based on an extensible elastic rod. They also parallelize movement of hairs to make the simulation more realistic. Here’s a link to the paper: https://graphics.pixar.com/library/CurlyHairA/paper.pdf
@carolyn-wang that is SUPER COOL that you brought up pixar- its exactly what I thought of for modeling hair, since I recall hearing that hair was one of the most difficult things to animate in early pixar (hence why the first movie, Toy Story, featured toys).
A related article/interview talks about how the hair in Monsters University was so detailed that it required massive farming and rendering jobs! https://venturebeat.com/games/the-insiders-view-of-the-tech-behind-pixars-monsters-university-interview/
I assume most hair is not a sequence of strands with material data, but rather a shape with a hair texture. How can you render light reflecting of hair correctly without representing each strand? My guess would have a pattern computed across all the hairs, then apply the texture to represent each hair.
I can't image how much time it would take to calculate all the different hairs, since each of them give off such different looks, and depending on each part of the hair, it's shiny, and the light passes through it. It's also interesting how the light catches each individual hair strand, and it's hard to simplify it since you can tell that each strand should react differently to the light
For Disney, with Rapunzel, her hair was represented with dynamic wires. This allowed different laws of physics to give it this more free form, flowing look. I thought it was interesting that they seemingly then just applied a hair texture and BRDF to these wires to further convince the eye this is hair (its shiny like the hair appearance here, differs in shading, etc). Furthermore, her head of hair was mapped as 147 tubes instead of animating around 140,000 strands of hair. The tubes themselves had a bunch of "breaking points" or "joints" that the tube would bend at to give animators more freedom in what form her hair took. I presume this also lessens the computational tax to render her hair and creates a more convincing finish.
One most simpliest way is to layer textures to create the illusion of fur. As furs are thin and transparent, it is always hard to render them correctly.