Zhiying Shi | 3D Character | Portrait of a Grandfather

Concept and Inspiration

My grandfather is a stoic man who doesn't like cameras — there is almost no photo of him around the house where he spends so much time taking care of. The inspiration for this project comes from my admiration for him and my desire to preserve a memory of him with my own skills and a combination of art and tech. The goal was to achieve a high level of realism, focusing on the intricate details of his face, hand, and the familiar setting of his study. This project not only pays homage to him but also marks a significant starting point in my journey as a 3D character artist.

Process Breakdown

Initial Idle Sculpt

The head sculpt began with a basic sphere until it reaches the primary shapes. Then I retopologized the mesh to create a clean, efficient topology.

With the retopologized mesh, I established the major forms and anatomical landmarks.

Using ZBrush layers, I added secondary shapes and fine details such as wrinkles and pores, and polypainted the skin texture in grey scale, which later became quite useful as I blend it with other texture maps. As non-destructive as I could keep it, this method allowed for continuous refinement.

Learning and Iteration

I wanted to call the sculpt finished, but NO — my knowledge of facial anatomy was very limited. Through many passes over the months, feedback from other artists, as well as studying expressions and the Facial Action Coding System (FACS), I improved the accuracy of the expressions and the anatomical details of the sculpt as I decided to turn this into a full scene.

Constant self-critiques and iterative adjustments were also very crucial in achieving the final version of the sculpt. I experimented with Fibermesh in Zbrush and thought it was enough before I finally teach myself how to use basic Xgen.

Scene and Look Development

I soon realized that the secret sauce in achieving greatest impact even through a single image is by always intending to tell a story. Clearly, a greyscale head sculpt and a simple mesh for shirt weren't close to enough.

More Action

I sculpted and updated the yawning expression using FACS, which involved activating specific Action Units (AUs) to depict the muscle movements accurately. More importantly, I asked other family members to take some more pictures of my grandfather so I could have better direct reference.

I sculpted a full arm following the primary>secondary>tertiary approach, then retopologized and posed. I used the hand to give a more specific intent to the yawning expression. Hand is arguably the second expressive part on the human body after the face.

I sculpted more folds and details on the shirt, breaking up the silhouette to give it a more natural look and guide the eye.

More Context

Modeled in Maya with several miscellaneous objects sourced from Sketchfab, the environment of my grandfather's study includes a Chinese calendar that he checks everyday, his calculator, radio, notebook, glasses, and many family photos — both framed and unframed or simply put under a piece of glass on top of the cabinet! Many eventually got blurred out but I enjoyed the process of creating these assets a lot, with their presence adding context and depth to his character.

More Details... and Passes!

Initially, I used FiberMesh as a placeholder for grooming. With XGen in Maya, I created the hair, eyebrows, and eyelashes again to achieve better shading. Throughout the lookdev process, I continuously adjusted the XGen groom to achieve a more realistic look as the expression itself kept being updated.

Countless passes in texturing and shading helped breathe more realism into the scene. I learned much more about controlling specular and translucency, especially around the cheek area where pores are usually heavy and the open mouth area where the translucency of the teeth and the moistness of the mouth meet.

More Control

To achieve a cleaner final render and more control over Depth of Field, I rendered the scene without DoF and used the Z-depth pass as a mask in compositing. This approach allowed me to apply a lens blur in Photoshop, faking the DoF and reducing noise significantly. This technique gave me greater flexibility in fine-tuning the final image and ensured a higher quality render without the artifacts that can sometimes accompany in-render DoF effects.

Conclusion and Future Goals

Through this piece, I gained invaluable experience in softwares, facial anatomy, expressions, and realistic character creation. The struggle was real but it also fueled my passion for character modeling. I am excited to undertake more projects in this field. My goal is to continue learning and creating, and I hope to return with more refined works for the Rookies Awards next year.

Thank you all for your time and attention!