Welcome back to our blog! I’m Coen van Camp and I’m responsible for most of the 3D modelling in RITE of ILK. This time around, we want to show you how we made our protagonists, Mokh and Tarh. I hope to make this post a fun read for fellow 3D artists and, of course, everyone else! I’ll try avoiding technical jibber jabber and hope you will enjoy reading through my process.
Character-modelling for a game is probably the most difficult form of 3D art there is! Especially when it concerns the protagonist. It’s the one 3d model (or in our case two) that you keep seeing during the entirety of the game – and for the most part, they’re the focal point of the game-camera! Besides that, you also want to create sympathetic and emotionally connectable protagonist(s).
We knew that before any modelling could start, we first had to define the characters in all aspects as much as possible, to ensure that everyone’s on the same page. We had to think about behavioral characteristics, colours, possible animations, story, background, etc. You do not (seriously!) want to make major changes to your character’s design in one way or another when your modeler is already halfway through modelling the thing. In my opinion, this is primarily caused by 3D modelling not being a very iterative process. If the team decides to make changes to i.e. clothing, that usually means throwing away old stuff and starting all over again. To counter this slightly, I split the character meshes in two: one to represent the character’s body and one for the clothing and hair. This way, we could minimize (not remove) the risk of throwing work on the body away in case anything major changed to the clothing or vice versa.
Image: Character concept art, made by Mau
Thankfully, I was provided with excellent art by our Concept Artist, Maurice Diemeer (Mau for short). I was also provided with a story-arc, which at this point in development was nailed down by Alanay. Afterward, the complete art team collected a ton of real-life references of tribal clothing and instruments.
When the time came to finally flesh out the characters, I definitely felt there was a good base of source material to start with! This type of confidence is essential. But before 3D modelling could really start, I still needed to do some research on children anatomy. I can call myself experienced with modelling characters, as I’ve modeled plenty. But in my history of creating humanoid characters, I’ve not once modeled children.
Finding reference images of child anatomy is… well, awkward, to say the least. I’m pretty sure that during my “research,” I’ve been flagged a couple of times by government agencies (‘sup, N.S.A, I’m talking about you).
Image: We collected moodboard images from the internet and do not own the rights
The thing is, you need references of their bodies (not just faces) to properly 3D model it. I needed to know in what ways a child’s body differs from a fully grown adult.
Image: In-development character sculpt (Tarh)
I definitely underestimated the difficulty of 3D modeling a kid. On the upside, I was able to cheat a little by spending less research and effort on areas of the body that were going to be covered up by clothing anyway.
I know I promised to keep the technical jibber away, but it’s only a small part, so forgive me…
We wanted to formally introduce Substance Painter into the project and decided to start using it for the characters. I realized that my texturing workflow was always a bit inconsistent and random, and in a way, Substance Painter helped in standardizing my workflow. A lot of what I learned during the texturing of the characters, including workflow improvements, I also applied to a variety of different texturing tasks afterwards.
A very cool feature of Substance Painter is the use of Generator Masks. It allows for very realistic looking weathering effects based on the details of your texture maps (i.e. normal map).
The process of creating both Mokh and Tarh went relatively smooth, and thanks to Substance Painter, I was able to speed up my texturing process a lot! However, there were definitely hurdles along the way and things I would tackle differently next time around.
Image: The Substance Painter Generator Mask in action on, well, literally a mask
Given the small differences for the characters, as Tarh is a girl and slightly older, I made Tarh slightly taller. Because there are physical differences, the characters couldn’t share the same animation rig. This made everything a lot more work for both the animator and Wouter (who worked on the animation state machine for both characters). To make things even worse, the height difference of the characters is hardly noticeable with our default, in-game camera angle.
Image: Mr. Red and Ms. Yellow
When the characters were finally implemented in-game, it solved a lot of our initial problems. We could finally remove Mr. Red and Ms. Yellow, our placeholder characters. With the new characters, players no longer get confused with what character they are playing mid-game. The differences in shape and silhouette of the new characters are much clearer. I also received a lot of positive feedback on how close my modelling resembles the concept art.
As for myself; I would have done things differently. After all, it was my first time formally using Substance Painter in a serious project. Since then, I’ve done a lot more work using the software and I feel much more confident with it. It also allowed for Substance Painter to be incorporated into the turtle pipeline. Regardless, I’m still quite proud of how the characters turned out. In my honest opinion, they turned out this well because we started out with a good pre-production phase. I hope you like it as much as we do! You can see the results below!