This post is part of our Art Pipeline Devtalk series, in which we break down the art production process, share our experiences in the past two years and recommend tips that worked for us. In our previous post we’ve started the path to asset creation by breaking down the concepting phase, and now we move to the next step – 3D modelling. I sat down with our own super talented Tomer Abadi to learn more about the process and share it forward:
We’ve left off our process after the art team got Liandra’s concept finalized, what happens next?
“After getting the final version of Lindara’s concept art which includes a render and an orthographic sheet, the modeling process could begin. My first step was to use Maya, the 3D software I chose to use, to process the orthogonal references along with a female base mesh model (which we created ahead of time as we knew our heroes will be female).
The modeling process of Lindara was a bit tricky as she should feel light, even while her robes give her tons of volume. It took some tweaking to develop an understanding of the right volumes and forms of her clothes, so that translating them from 2D concept to a decent 3D model will maintain that ‘light’ feeling along with rounded cloth shapes.”
How do you polish and iterate your model beyond the initial conversion?
“Personally when modeling I break the work process to smaller chunks: in Liandra’s case the first step was modeling a quick proxy version with all of her props, clothes, robes and carried items. After seeing that all of the elements and shapes fit together in a pleasing way that conforms to the concept I received, I passed it over for initial review of the art team.
In general, it’s important to let fresh eyes take a look at your work, seeing as after a while modeling your own eyes lose objectivity and stop noticing (sometimes obvious) flaws. With Liandra, every review brought a slew of great feedback which helped me tweak and improve her form to maintain her shape and her feel while becoming more and more convincing when it comes to motion and physics.
For example, I was messing around with Liandra’s hair for hours, as she’s designed with a bit of hair falling on her forehead and getting the volume and shape right was challenging. It’s pretty common for a great 2D piece of concept art to look bad when converted to 3D, but surprisingly in our mage’s case it converted well and took fewer iterations than expected.”
Could you give an example of how decision making in the modeling process affects other aspects of development?
“You mean, except for deciding how our heroes, enemies, environments and objects look like? 🙂
Well, there are many ways to model, and especially when it comes to UV unwraps I have several choices to make that affect both code and other artists down the production line.
For example, Liandra’s skirt – we had a choice between unwrapping and placing the skirt-pieces one on top of the other, which will decrease variation but will save texturing time by mirroring the textures, while also helping us get more details into the skirt as a whole.
Our other option was separating the pieces and placing them next to one another in the UV map, which will increase variation across the skirts texture but will force smaller resolution available for each piece.
Eventually we decided on a middle ground by doing small pieces together and separating bigger pieces that require more detail and variation. Once done, our mage was ready to go forth into texturing.”