My final Capstone project involved surfacing and grooming a character bust. The main purpose of this project was to explore stylization in skin shading, groom shape language, and approaches to Look Development that I had never previously exposed myself to. 
ACKNOWLEDGEMENTS
Before I move on with the technicalities of the project, I would like to acknowledge the people who have been instrumental to my project! 

First off, thank you to Yongdan Zeng for providing me with a beautiful character model to surface.

Thank you also to Megan Walker, Dave Walvoord, and Caleb Kicklighter who served as my mentors throughout the process. Their expertise meant that I learned so much.

I'd also love to give a final thank you to my peers in Visualization, specifically Julia Boisvert, Alyssa Curran, Jenna Aguilar, Gabby Gonzelz, and Daniel Arellano, all of whom surfaced characters alongside me this semester.
REFERENCES & RESEARCH
I began my project by simply gathering references from the internet, but I quickly realized I couldn't always find the exact high detail photos that were needed to understand certain aspects of light interaction with skin. My mentor, Megan Walker, suggested me and a group of a few more artists take our own photos for reference, with a light set up we could easily replicate in 3D. The above images were just some of the results from that photoshoot. Using one singular light source meant that we could get a clear picture of the terminator on the human face, which would allow us to better develop the Subsurface Scattering aspects of our skin shading set up.

We also explored the cross polarizing technique to remove the specular from our images. That way, we could view only what our albedo maps should look like. The following images were some results from that shoot.
Beyond gathering my own reference, I researched how other artists had achieved rendering skin in RenderMan, peach fuzz grooming, and subsurface scattering shading.

While they were great resources for understanding the fundamentals of skin shader development, they were all aimed at creating a completely realistic product. I couldn't find any extensive resources for stylized skin shading, which is what I explored through this project.
THE PROCESS

Subsurface Scattering
With the photoshoot references at hand, I was able to create the same light set up in my scene, and I began testing different SSS mean free path distance values. SSS needed to be developed first, so that all the other aspects of the shader could follow.
Based off of these SSS wedges,  it was determined the MFPD range should fall somewhere between 5 and 6. This led to more wedges, but this time with a minimum and maximum MFPD value. This was achieved by creating a map to describe the varying MFPD values for the model, and then remapping those values.
The image to the left was derived from the Thickness map generated in Substance, with some slight paintovers in the chin because it was too dark. The thickness map describes areas that are the most thin with a black value, and areas that are the most thick with a white value. My initial instinct was to think that inverting the map would give you the correct results for your MFPD in RenderMan because there is more scattering in thin areas of skin. However, doing so means that the MFPD becomes so high in thin areas that they begin to lose edge information. To preserve that detail, the map remains like the one pictured.
Using a remap node allows us to unclamp the output so that the output values are not restricted to values from 1.0 to 0.0. With this in mind, the output values can describe the MFPD at its minimum where the map is black and at the maximum where the map is white. I performed another wedge test changing my minimum and maximum values for the MFPD.

It was determined that the SSS values for this particular model should be, at minimum, 6, and, at maximum, 15. The result is pictured below.
Roughness & Displacement
The next portion to address was the roughness and displacement. Some progress renders have been attached below to show what each part adds to the overall look.
Displacement was achieved using pore maps, and then backing off on high frequency detail due to stylization. The specular on the shading set up, on the other hand, needed an extra component. 

Human skin is composed of an oil layer, which creates a separate specular reflection from the base skin. The need for two specular lobes meant the addition of a Clear Coat, which was a map that was painted separately in Substance Painter.

The oil layer map is painted white where the skin gets the most oily, but certain areas are painted black in order to better control the art direction of the piece. For instance, a sharp specular hit beneath the nose might make the character appear like he or she has a cold. The oil layer map I used is pictured below, alongside an image depicting with how the PxrSurface shader deals with Clear Coat.
Color Map Information
Color was the last component of the actual shading set up for the skin that needed to be developed. The main challenge with color was ensuring the specular did not blow out the lighter parts of my color map for this skin tone. There are also a lot of high frequency details in realistic skin that had to be blended out in this color map.

For the skin tone I was going after, a satured salmon worked nicely as the SSS color. The process of combining all the texture maps we've discussed so far can be viewed below.
Eye Shading
In order to create an eye that reacted to light in the correct way, the eye needed to be modeled using a few different meshes. The model was split into the Cornea/Sclera, Iris, and Pupil, all with their own shading setups.
The Cornea/Sclera mesh was shaded with a PxrLayerSurface, with one part describing the Sclera and the other describing the Cornea. The Sclera layer is just a white subsurface material, while the Cornea layer is a glass material. The shading network can be viewed below.
The contact between light and a human iris creates caustic reflections, but rendering caustics can be costly. It can also be difficult to fine tune the caustic effect, which is why I took the approach of faking the look. This meant that the iris was rendered with metallic properties rather than diffuse properties.

The pupil is not meant to interact with light at all, so a PxrBlack material, which has no specular, was assigned to it.
Eyebrow & Eyelash Grooming
The eyebrows were created by placing XGen guides and shaping them in the direction that the human eyebrow flows. The biggest challenge here was to create an eyebrow that subtly broke its silhouette so that it looked natural, but also looked appealing. The image below describes how the guides were shaped.
The eyelashes, much like the eyebrows, needed to look natural, yet maintain a graphic and appealing shape. In order to do this, I modeled a geometry that described the silhouette of the eyelashes. I selected some of the edges from that model, and I converted them to curves. From there, the curves could be shaped independently using CV's. Once the curves were shaped how I wanted them to be, I converted those curves to guides with XGen's Curves to Guides utility.
From there, I added modifers to both, and the final results can be viewed below.
Peach Fuzz Grooming
Peach fuzz grooming proved a harder challenge than I expected, since the effect needed to be a very subtle one. The fuzz was created using groomable splines, which were combed to follow the directionality of the model. 

The hair primitives needed to be much thinner than the other hair in my scene, and the shading proved to be essential for creating the right look. My first light spin render revealed that the peach fuzz was catching a sharp specular at certain angles, making her face appear unnaturally hairy. This shading problem was solved by changing the color of the hairs to white, and shifting the Azimuthal Roughness all the way to 1. These changes reduced that shine and created a much more subtle effect.

Character Lighting
The V-Dev rig lighting was creating really harsh shadows on the character's face, which was fine for diagnostic lighting. However, the lighting needed to be altered for a beauty shot. 
The light set up is pictured to the left. It consists of the following elements:
The edge light serves as a way to catch the edges of the hair and cheek on screen left.
The key light is the main source of light illuminating most of the face.
The fill light serves to fill the shadows created by the key light. This light is slighly warmer to add some dimensionality to the image.
The other fill is positioned behind the character and serves to catch a highlight on the parts of the hair on screen right. This light is also slighly warmer.
There are two sphere lights in the scene as well which serve as character eye highlights (one for each eye). These are light linked to the Cornea of each respective eye, so that they don't effect anything else in scene.
Key
Key
Fill
Fill
Edge
Edge
Second Fill
Second Fill
FINAL THOUGHTS
This project served as an incredible way to introduce me to a lot of new techniques regarding Look Development. It also taught me a lot about the stylization of forms and surfaces, in order to create a more appealing character. There were many The process was integral to my development as an artist, and I'm excited to take these principles and skills I've learned and to apply them to a new challenge.
↑Back to Top