Simon Le Paih is a Technical Artist working on the Substance 3D Assets team. Here, he discusses how the team tests out the practical uses of the Substance 3D content, and the choices and processes behind his robot arm scene.
The origin of this project was the release of over 2000 meshes on the Substance 3D Assets platform. A great deal of research and background study goes into this sort of release, and that was a large part of the motivation here – this range of meshes had been made available on the platform, but we wanted to stress-test them; we hoped to learn more about how valuable and useful these assets can be to artists in a real-world project, and the precise ways in which artists will typically use them.
And so a few internal artists were given the pretty open brief of creating whatever artwork we wanted, using meshes available on 3D Assets. Essentially, we were asked: what can you do to showcase the 3D Assets meshes, and the potential for kitbashing that they provide?
Final render in Substance 3D Stager.
One key point that I wanted to transmit with this project is that the Substance 3D resources are usable just fine exactly as they are – but they’re also part of a wider ecosystem. When you take advantage of the interconnectedness of the various Substance apps and resources, you find that you have an immense toolbox that allows you to do essentially anything you want. I really wanted to highlight that idea of the Substance 3D ecosystem as a toolbox, rather than simply picking out some elements, doing some staging, and getting an attractive render.
The robot arm scene in Substance 3D Stager.
At that time, the 3D Assets platform didn’t have a section dedicated to game art, or to sci-fi (though we’ve added a sci-fi collection since), so I took it as a personal challenge to create something in this area. That fitted well with my background in video games, too.
Breaking up, and getting back together
Each artist took his or her own approach to the meshes available. My own preference was to break down the meshes available into their component parts, to see if I could reconstitute something different, that fitted my chosen theme. I had a good idea that I wouldn’t only use Adobe software, but I wanted to remain within the Adobe ecosystem as much as possible.
Above all, I wanted to create ‘clean’ renders, as time-efficiently as possible.
I also specifically didn’t want to use any concept art for this project, precisely because this sort of process can be useful for creating concept art. Concept artists increasingly need at least a minimum understanding of 3D workflows, either because they’re creating concept art for 3D artists, or because it’s practical for them to use 3D as part of their own workflow. One potential use for the meshes in the Substance 3D Assets library is to help concept artists, who don’t necessarily have any experience in 3D modeling, to assemble a scene. With the robot arm project, I haven’t done any modeling in the classic sense; I’ve simply cut up and reassembled different pieces. The robot arm is a finished image now, yes – but it’s also kind of a concept piece. It’s an exploration of shape.
But I’m getting ahead of myself. The next step was to go through the 3D asset library and pick out some meshes I thought might fit my theme. I selected maybe 20 or 30 meshes, in total. I then broke them down into their component parts in Blender, and reassembled them in new ways. I created a few possibilities for the robot arm, like this. Then I collapsed their meshes and cleaned their UVs.
I’ll note that one of the great joys of this project was this phase of browsing through all the assets on the Substance 3D platform, and breaking up and reassembling objects. I really enjoyed this kitbashing approach to creating a hero asset. Rapidly seeing that you’ve created a viable robot arm like this – in maybe just 15 or 20 minutes – was super-interesting.
Once I was satisfied by some shapes for my robotic arm, I sent them to Substance 3D Painter to carry out some quick tests with a few Smart Materials.
I ultimately chose one arm in particular; this is the one that appears in the final scene.
Setting the scene
I still needed to create a scene around the robot arm. I chose a sort of ‘automated generator’, with my robot arm transporting an energy cell from a dispenser, and placing it within the generator itself.
I also used Substance 3D assets to create the scene around the arm, though I used some of them in ways quite different to their intention as stated in the asset library. For instance, the screen that’s present was originally just a panel; the mesh wasn’t intended to be a lit screen.
Most of the assets in this scene were suitable right out of the box. I just made some small modifications as needed. For instance, the only modification I made to the generator was to create an extrusion from the center; the arm places the energy cell inside this space.
I’d previously had to redo the UVs for the robotic arm, because it was completely composed of parts broken down from other meshes. But for everything else in the scene, the UVs were clean. I could just texture them directly in Painter. This was much faster than doing everything from scratch.
Test renders in Stager, and texturing
Test renders in Substance 3D Stager.
I used a couple of different rendering engines in this project. Quite early on, while I was still in the conception stage, I used Substance 3D Stager to do some test renders, to check lighting and color. I opted for Stager at this point purely to save time, and to keep my workflow simple – Stager was quick to use, with few constraints; it allowed me to create some draft renders fast, and it also provides some maps and masks which I used to do some post-processing in Photoshop. Creating some samples of lighting and color tests like this helped me to make choices concerning the texturing.
I textured the scene with Smart Materials, making some modifications as seemed most appropriate. Then I baked my scene all at once, with each mesh having its own texture set. I find this is a quick and straightforward way to get results quickly, and keep everything consistent.
I imported all my maps into Substance 3D Designer to make .sbsar files with parameters allowing me to tweak roughness, base color, and emissive.
More renders – Substance 3D Stager and Unreal Engine 5
For beauty renders, I used both Unreal Engine 5 and Stager, depending on the context. For the close-up renders I used Stager, in order to have cleaner images. My previous reasons for using Stager – speed, simplicity of workflow – were big motivations here, too. Stager’s interface is very user-friendly; you can assemble a scene very easily. The lighting is very straightforward and, as mentioned previously, the renders come with a certain number of maps, which again made it very easy to add some post-processing effects, such as blurring due to distance, or around some light sources. Being able to add these effects in post-processing, rather than creating them in the render itself, was a big advantage. If you add blur to a render, for instance, and then you find it doesn’t look good, you have to do the whole render again. But if you add that effect in post-processing, you can fine-tune it much more easily.
Another good point with Stager is that you can send your render directly to Photoshop, along with all your masks. That takes care of half the workload involved in the post-processing phase, immediately.
Final render in Substance 3D Stager.
For the ‘factory images,’ featuring a few different arms, I used UE5. There was a creative reason and a more technical reason for this choice. Creatively, I wanted to test out the scene in a game engine. And that’s very much because my background in games made this interesting for me, and because it fitted in with my original motivation of wanting to create something with a sci-fi/game feeling.
First lighting pass.
More technically, I notably wanted to test out Nanite, which allows you to more or less bypass polygon limits; this gave me a lot of freedom to enjoy myself with the scene’s lighting, and a few other aspects besides. I also wanted to add some visual effects available in UE5, such as particle systems, ways to improve the lighting, and fog. I used UE5’s Cable Actor function to simulate the physics of the cables in the scene, and I applied some decals from the Substance 3D library in UE5.
Material integration and Cable Actor creation.
Adding decals from 3D Assets library.
Final render in UE5.
Final render in UE5.
The assets and tools available in the Substance 3D Assets library open up a wealth of creative possibilities in areas such as concept art, 3D scene creation, illustration, virtual photography, and more – even to people who aren’t necessarily experts in 3D.