Elvira Trofimova told us about the production process behind the FLCL Crew project, talked about the process of character creation in Maya and SP, and discussed her student years at Gnomon School.
Introduction
Hi, My name is El (short for Elvira) and I’m a Character Artist. Originally from Russia, studied there at Samara State Academy of Culture and Arts. After graduating, I worked on social and mobile games, one of the most memorable – Cut The Rope by ZeptoLab in Moscow. Moved with my husband to New York in 2014, worked there as a Concept Artist for a collectible figurines company, and then decided to switch to 3D.
I went to Gnomon School in California and after graduation in 2020 got hired by Treyarch to work as a Character Artist on Call Of Duty: Black Ops Cold War. Just recently finished my one-year contract with Treyarch and am currently working on some personal projects.
Studying at Gnomon
When I was working on concepts for collectible figurines I started using 3D for blockout and posing. I took Intro to Zbrush class at CGMA with Michael Pavlovich and started implementing 3D into my personal and freelance work as well. In 2017 I and my husband went to Los Angeles for E3 Expo and got a chance to visit Gnomon for an Open House event. I was blown away by the atmosphere and level of student work presented there, stats were impressive too – very high rates of employment upon graduation. So we took a leap of faith and I decided to apply to Gnomon’s Game Art Track – a 2-year intensive program with a concentration on real-time production. 6 terms with 6 classes each term, plus the last two terms dedicated mostly to Demo Reel.
The FLCL Crew Project
During my last term at Gnomon, I took Miguel Ortega’s Look Development class as my elective to learn more about composting, rendering tricks, and, well, Look Dev in general. It’s not a modeling class and I had most of the assets done previously in earlier classes – Rickenbacker Bass I modeled before Gnomon for Michael Pavlovich’s Intro to Zbrush and textured for Lighting and Rendering with Oded Raz. Vespa and Canti (scooter and robot) were made for Hard Surface 2 with Max Dayan, and Haruko’s base model was made for Digital Sculpting with James Schauf.
You could have guessed by now that I’m a big fan of FLCL so whenever opportunities for fan art projects came up I took them! I just had to make adjustments to the final composition, update models and textures and add the environment.
My initial pitch looked like this:
Then, I scrolled through the show to pick an environment for my characters, here are some of the options I had:
I chose the “bike on the curb” variant and quickly compiled composition in Maya:
Creating Characters
Haruko was created the old-fashioned way – starting with ZSpheres in ZBrush and retopology in Maya. I didn’t try to push proportions or anything like that, her level of stylization was something that came naturally for me at that point:
Her clothing was made in Marvelous Designer:
I used Mixamo for a quick rig and pose and tweaked the pose manually in Maya. With that same pose imported in Marvelous Designer as a Morph Target I simulated the clothing:
For her face textures, I used texturing.xyz’s Female Multichannel Face #07. I placed Texture Maps over her UV Map and with Photoshop’s liquify tool quickly warped textures to match her face’s UV Map. I knew she was going to be rendered at a medium resolution so I didn’t need the highest fidelity in this case.
I made her hair with Xgen (Core) – the most challenging part here was the balance between stylization and realism.
Canti, on the other hand, was made completely in Maya using reference from the FLCL artbook.
Block-in:
Jacket pattern in Marvelous Designer:
Final topology:
Scooter and Guitar
For Vespa, we had to analyze the shapes and plan ahead before jumping into modeling.
Haruko has Piaggio Vespa Super Sport 180 but a blueprint for this particular model was hard to find, so I ended up using a similar model’s blueprint – Sprint 125 and side view or Haruko’s Vespa from FLCL Artbook.
The same process as Canti, I used mostly Modified Primitives and Loft Curves, going from big to medium to small shapes;
I made Rickenbacker Bass before I knew how to use Maya, so I used my old model as a blockout and re-made it from the ground up:
Haruko has left-handed Rickenbacker 4001. Always important to do proper research! As a fan of FLCL, I have a few artbooks that I used extensively for those projects, especially The FLCL Archives that has lots of character designs and blueprints, very convenient to use for modeling!
Texturing
I used Substance Painter for all the texture work, here’s one of the assets I made in Props and Weapons for Games class with Chris Stone:
Similar to modeling we start with block-in. In the case of texturing we block-in color and base parameters – usually Roughness and Metal for real-time assets or Spec and Gloss for render assets. Then gradually layering variations on top – color variations, dirt, wear, scratches, etc. The most challenging part for me in texturing is to make sure the materials don’t lose details on the rendering stage. We constantly rotate model and lighting in Substance Painter and are able to see how light reacts with the model dynamically, but when rendering static images with different lighting setups I often end up losing those subtle details so I try to emphasize them and check periodically by rendering test shots.
Lighting and Rendering
I wanted that “Deer in the headlights” photo-effect like someone was driving and suddenly saw Haruko and Canti on the road shoulder. So I tried to place the main light source like it was a car’s headlight. One of the coolest tricks I learned in Look Dev class is that we don’t have to animate light in Maya and render animation sequences! You can render each light as a separate source and animate it in post-production. It requires a little additional setup in V-Ray Render settings:
Add LightSelect render element and drag the light source into that node in the outliner.
I rendered a single EXR image with all the render passes as channels and animated each light source channel separately. Additionally, I rendered blades of grass on the foreground for a parallax effect and animated a slight camera zoom in Nuke.
Conclusion
It is hard to compile all the knowledge from 2 years of intense studies at Gnomon into a couple of sentences but I’ll try to remember some – definitely, the trick with rendering Light Select in V-Ray described above. Another cool trick I learned from that class – color matching by channels, description can be found here. Another one from Photoshop class – Frequency separation – a very cool technique for photo retouch. There were many more, of course, those are just some that came to mind.
One of the many important things that Gnomon taught me was discipline – having a schedule and deadlines made the process tight and structured. It was stressful, yes, but for me personally – very rewarding. Instructors were always there to help and guide students whenever we had troubles. I learned that there are no stupid questions and I asked A LOT.
And that’s my advice for beginners – don’t be afraid to learn new things and never be shy to ask questions. Just do it!
In conclusion, I’d like to thank all my Gnomon instructors, especially Miguel Ortega and Anton Napierala for their guidance through this project; my husband Alex for supporting me through my Gnomon journey, our cat Musya for her emotional support and, of course, 80 Level for the opportunity to look back on this project and share the process with you. Feel free to contact me via ArtStation or Instagram.
Elvira Trofimova, 3D Character Artist
Interview conducted by Theodore Nikitin
Keep reading
You may find these articles interesting