By design, they should visually be descriptive, unique, and feel collectible. And if you ask production they should be finished in two weeks, as we had to time constrain ourselves to a single sprint for this feature.
Now that we had proof we could make this it was time to make some creative decisions. As our workflow was procedural and semi-automated we didn’t really have to make them right away but we wanted to experiment and set a clear path as early as possible. Iteration was super easy, barely an inconvenience. Thanks to the PBR node we could just plug in any HDRI image and immediately get a different light, rig, and mood. Alongside HDRIs we also tried building completely custom light rigs using the Substance Designer Light nodes but we opted for the HDRI approach as it was faster. In the end – the decision came to choosing the light setup which was most neutral and readable.
The last thing we were uncertain of was how to combine the primary and secondary elements of the badge. Yes, there was a clear option of doing that in Substance Designer but at that point, we didn’t have the process of generating heightmaps nailed down. On top of that, we didn’t know what those elements would look like, how big they would be – aka what their height ranges would be. They might overlap in some way and we didn’t want to build systems to optimize and balance all of that. We went a bit old-school and just combined the elements in Photoshop. It was rather simple to build a Layer Style that would match the shadows of the PBR Render node.
Once the models started coming in we were extremely happy with the results and whoever saw these first outputs didn’t believe us that we haven’t “properly” rendered them. Victory!
For the secondary (background) elements we went all in and used everything we had – procedural, alpha library, some custom made ones and we even made a quick shield generator in Substance Designer that was used in more than half of them.