Avyanna: Working on an Animated Stylized Character

Introduction

Hi! I'm Marc Martin, I'm a 3D generalist who's working in a Spanish animation studio called 23Lunes. I also love creating my own characters and giving them a backstory. I just finished my studies in La Salle Barcelona, but the reason for the knowledge I have is because I'm a very curious person and I learned a lot of stuff on my own and from other artists.

Almost all the projects I've worked on are personal projects, it's only now that I'm starting in the 3D industry as a professional. In the last eight months, I've collaborated in a publicity spot for "Cuetara" and a short called "Herman".

My passion for creating characters started with watching Disney movies when I was a child. It started as a hobby because I always loved drawing and then I discovered the 3D world.

Avyanna: About the Project

It started as a school project but I developed it as a personal one. I worked on it in a team that helped me with all the ideas I wanted for the character. My goal was to improve and challenge myself as an artist, that's why I wanted to try to put as much love into this project as I could. But that's just the beginning because our idea right now is to get this project into a teaser. We've tried to create our own style relying on the references from Overwatch and Riot's cinematics, mixing them with some cartoon animations made by Disney or Pixar. With that, we've reached a style that gives cartoon vibes but with realistic shading. Also, we want to match it in the teaser with 2D animation.

Sculpting

We've worked without a concept and went for very stylized forms with cartoon proportions. We worked for 3 weeks just on the face and the body. The work has been a constant back and forth process during which we looked for simple shapes. One of the biggest inspirations was Overwatch's and Disney's faces and proportions.

I always use ZBrush for sculpting characters. I think that is a super artistic tool in which you don't need to worry about any technical aspects. The brushes that I use the most are Move to get basic shapes, ClayBuildup to figure out volumes and muscles (anatomy) and, at the end of the sculpting process, I use DamStandard to make more details.  

Working on the Outfit

We haven't worked with Marvelous before, so we have had to do a lot of research to get a good result.

For the teaser, our intention was to get the result without losing the essence that Avyanna has right now. The real problem appeared when we had to simulate the clothes; we did not get the result we were looking for and we had to sculpt frame by frame!

The process was to take out the Marvelous simulation, go to Maya and make a retopology that really worked for us. From there, we got a proxy and used it for the simulation within the animation.

Retopology and Unwrapping

The retopology stage in Maya was very important to us and we worked hand-in-hand with rigging to get the best possible deformations; we tried to make different loops follow the musculature to achieve more natural deformations. Again, we were very inspired by the faces from Disney and our intention was to achieve facial deformations in animation as similar to that style as possible. The body was interpreted a little more according to our needs!

Having a good topology is crucial for the animation because it facilitates the subsequent processes such as the unwrapping of UVs or the painting of weights (skinning).

The clothing was easier since we were going to simulate it, and we did not have to worry about the rigging deformations. The only thing we were concerned about was following the seams in Marvelous!

As a piece of advice, I would suggest not being afraid to copy a well-done retopology. This is how you learn, by observing and practicing. Also, pay attention to something known as the "flow"; it means that all the loops should be equidistant from each other!  

Texturing

The texturing of the skin was a challenge since we have never done something as complex and were not used to that level. The process took place entirely in Substance Painter. The skin was made with photoscans purchased from the XYZ website; this gave us a very good result!

Then in Arnold, we set up the final look. We used a fairly complex shading playing with all the layers (SSS, Coat, Spec, Sheen, etc.) and masks to have more control within all the properties of the AiStandard.

Within Substance Painter, we didn't use many automatic masks. We used some as a base tried to create all the details manually to have more control and make everything look more natural!

Grooming

For grooming, we have used Xgen within Maya. For the fur on the hat, we used two collections: one that provided us with the base and another that gave us long hairs. For the eyebrows and eyelashes, we used curves to create directionality.

Rigging and Animation

The body rig was built using the autorig for Maya our rigger is developing and scripting. The face is a mix of the first layer of deformation using a blendShape system and a joint-based rig.

When you animate such stylized characters, you must find the attitude of the character and manage to capture that in their expressions as well as identify what their most extreme poses can be. Depending on the style that you want to get, you should look for very exaggerated or much slower movements; this will also vary depending on the emotion you want to transmit. For example, for Avyana, we have faster transitions when she is happy or surprised; when she is sad, to reinforce the idea of her expression, we make a slow transition to capture the details of the facial expression well.

One of the main challenges that I found when animating Avyanna was to correct the simulation of the clothes since we did not get the results we were looking for. The clothes looked a bit floaty, stiff, bounced too much, and did not match the material we had in mind for the jacket. That's why we made the decision to sculpt all these rebounds and almost frame the simulation to make it look just as we wanted.

Rendering

We used a Render Farm, with the Arnold Engine by SolidAngle. It took an average of 50 minutes to render one frame.

The lighting is a neutral light rig, but the bounce light is warm. We focused on every expression and played with colors according to the emotion. For example, with sad expressions, the light is cold and for happy ones, the light is warm. For other expressions, we focused on the intensity using 2 key lights in the setup.

We modeled the tears and then lit them apart using only the specular option. In this scene, we wanted to focus on the emotions so we highlighted the silhouette using two rim lights with different colors and placed the key light on top of the character for a bit of drama.

Marc Martin, Character Artist

Interview conducted by Arti Sergeev