Chung Jui Lee shared his approach to recreating the main character from the Peaky Blinders series during the CGMA course and discussed his skin creation workflow in ZBrush and Substance Painter, and hair generation in XGen.
Hello, my name is Chung Jui Lee. I’m currently working as a 3D Artist at IGG Inc. in California, United States. I’ve been working here for 2 years after I graduated from Savannah College of Art and Design.
I always loved to draw characters from a young age. At first, I tried Industrial Design during college, but most of the products I made turned out to be too cartoonish and character-like. Then I came to the U.S. to learn animation at SCAD and I’ve been modeling characters ever since.
I love learning new stuff, it helps me stay motivated and challenged. And I kept seeing lots of great artworks from the past students of this CGMA course. That’s when I decided to enroll and see what I can come up with.
Before this class, my portfolio mostly consisted of stylized characters. So, I thought it would be beneficial to add a realistic character to broaden my scope. My goals for this class were to understand the workflow of realistic character creation and to learn different software and tools along the way.
About the Character
Peaky Blinders was one of my favorite TV series, I love its aesthetic and the character of Thomas Shelby. And I thought Cillian Murphy has a very unique face that would be interesting to sculpt.
For me, this is a perfect subject for this course. The overall character is relatively clean and simple. I would be able to focus on pushing the quality and finishing it in 10 weeks.
For the reference, I used mostly from the show and some other amazing 3D arts as well for my quality goal.
I had a human base mesh that I can start sculpting from. After establishing the basic anatomy, I then exaggerated his distinctive features, like the squarish jaw, pronounced cheekbones, and his deep-set eyes. It’s always a good idea to start as a caricature first, then slowly reduce the intensity of exaggeration. I think this way is very effective to get the personality quickly.
Another important thing is to not go into details too fast. Focusing on the big shapes, proportion, and personality first. It will pay off in the long run.
For the clothes, I used Marvelous Designer with some detailing in ZBrush. The geometry is quite simple, just a couple loops around the parts that will deform and keep the rest straight and clean.
Before this class, I thought it would be time-consuming to produce realistic skin textures. But it was actually very straightforward and fast.
Here is my displacement map creation process: 1. Prepare the head mesh with efficient UVs → 2. Find the proper skin textures from Texturing XYZ → 3. Project the high-resolution maps onto the head mesh using Wrap 3D → 4. Clean up in Mudbox and Photoshop → 5. Sculpt additional details in ZBrush.
The same goes for color maps, I used a female head texture from the 3D Scan Store and overlaid the displacement map on top to match the detail. With some color adjustment and cleanup, I was able to produce high-quality skin textures in no time.
Creating Fabrics and Clothes Details
My process for the fabric textures: 1. Block out the color first to see the big picture → 2. Find appropriate fabric materials from Substance Source then modify it to my liking → 3. Procedurally create some damage and discoloration with the baked curvature and AO maps → 4. Manually paint final details to break up the procedural patterns → 5. Add fuzz with Xgen.
For the buttons, tie clip, and the pocket watch, I did the polygon modeling with Maya and used Substance Painter for texturing.
The hair was actually the most challenging part of the whole course for me! Even though Pete did an amazing demo on XGen interactive grooming, I had a hard time matching the hairstyle. I ended up using XGen core for the hair, I like its non-destructive workflow and the ability to control the clumping with guides.
My first step was placing the guides for the hair to follow. Then, I painted a region map to divide the long hair (top) and shaved hair (back and sides). I also had a density map for a smoother border transition. Once things were set up, I added the modifiers as the final step to shape the hair further. I used clump, cut, and noise modifiers for this case and just tried to match the reference as close as I could. I highly recommend Tom Newbury’s in-depth Xgen tutorial if anyone is interested.
As for the lighting, I always tried to find some good reference for outdoor, indoor, and studio lighting first. So, I would have a rough idea of what I want. I also did several HDRI lighting tests (Images are from HDRI Haven). For those, I just kept the SkyDome intensity as 1 without any other lights in the scene.
Once I found the HDRI I like, I added the 3 point lights system for fine-tuning. But I also tried to keep them integrated with the background. The key light is for sculpting the forms and emphasizing the character’s features. The rim light is for outlining the silhouette, hair, and also bringing out the SSS of the ears. Finally, I create some volumetric fog to add a bit of mood.
The biggest challenge is the entire workflow. I have to go back and forth with different software, and it can be very time-consuming. But Pete made it very organized and step-by-step, I understood the workflow quickly and benefited a lot from this iterative process. Also, I learned to be more patient for those dull but essential steps. Once I overcame those obstacles, the result was very rewarding and gave me a sense of accomplishment.
For my next project, I want to create a real-time character using the techniques and concepts I learned from this course. It would be fun!
Thanks for reading! Feel free to reach me at ArtStation or LinkedIn.