My name is Mingshun Zhu. I am a recent graduate from Michigan State University, and I am working as a 3D character artist.
I started doing 3D modeling back in 2014, and I am mostly self-taught. My obsession with photorealism in storytelling leads me toward film production. Five months ago, I enrolled in a course on CGMA called Character Creation for Film and Cinematics by Pete Zoppi, and I’d like to talk about the things I learned from this course.
I decided to create a character that is simple in design, so I have extra time to learn and digest. X-23 from the movie Logan was perfect for a short-term study because of the ideal complexation.
Start by gathering as many references as I can. I then filtered out most of the images and kept the one that has the best demonstration of features.
Instead of going straight into the hi-poly sculpting or clothes simulation, I took the advice from Pete to get the base mesh done first. Making low-poly base mesh helps to define the scale and topology of the model for both referencing and sculpting purposes.
Continue with face sculpting in ZBrush. I prefer to achieve the correct facial anatomy before I get into likeness sculpting. Here are a few tips I learned from Pete. While sculpting the face, don’t push too hard and don’t admit too fast; subtle changes make big differences; keep refining and iterating the face in different angles and spots, and avoid adding too much expression on the face. Also, be very cautious of the age, gender, and weight to get the facial features right. For example, little girls tend to have a rounder chin and jaws and are wider in the eyes.
Clothes are generated using Marvelous Designer (MD). I used the base mesh I created earlier to get a rough layout of panels.
Be thoughtful of the relationship between folds and animation. Depending on the pipeline, inappropriate folds might cause deformation issue in animation.
After the simulation, you can either choose to retopologize the mesh using the Zremesher function in ZBrush or manually resurface the mesh. In my case, I used Zremesher to get the base topology done and then build on top of it using Zmodeler.
Before you start texturing, it might be helpful to think about the type of shot that the character is being used in. A close-up shot may require more detail on the character than a medium shot does.
I used various scans from TexturingXYZ and Surface Mimic for skin and cloth projection in Mudbox. Photo scans grant you fast and good results, but there might be no transition between sculpted and projected details. Thus, you may need to add extra wrinkles and folds to tie everything together.
The skin shader is a combination of Vray Fast SSS2 skin shader and reflective material. The additional reflective material helps to control the oil, wetness of the skin.
Clothes shader is very straightforward but with a little twist. At a glancing angle, the cloth looks brighter, softer than its base color. I can fake the desired look by blending the base material with a smooth edge material. SamplerInfo node determines which part of the cloth is facing the camera and which part of the mesh is facing away from the camera.
Hair was done using XGen in Maya. Reference is the key. I set up the hair in three layers. The first layer encompasses the style and shape of the hair. The second layer has less volume and higher noise value. The third layer only has a few strands of hair, and I manually placed them in the desired places. The second and third layer helps to make hair looks natural and lively, but it is incredibly challenging to get right.
I hope this article is helpful to you. I am glad to take the course because it helped me to speed up the learning process. If you are interested in learning more about character creation for films, you can consider taking the Character Creation for Film and Cinematics by Pete Zoppi on CGMA.