Fabric Collection Made With Substance 3D: Three Different Approaches

Introduction

Hi, I'm Elouan Harmand, and I've been working as a Material Artist in the video game industry for the past seven years.

My journey began with a master's degree from Supinfogame in France, where I learned the fundamentals of game creation. During my studies, I had the incredible opportunity to intern at Allegorithmic (now Substance 3D), where I learned to use Substance directly from the creators of the software.

After graduating, I moved to the UK and worked at Sony London, Rebellion, and Build a Rocket Boy, progressing from Material Artist to Senior Material Artist. Some notable projects I've contributed to include Blood and Truth, Evil Genius 2, Sniper Elite 5, Mindeye, and Everywhere.

The Fabric Collection

The Fabric Collection project was a way to encapsulate my past experiences. I didn't limit myself to a single technique but instead showcased a variety of methods to highlight their unique advantages. This approach allowed me to explore and leverage the best aspects of each technique, inspired by my previous work and experiences.

Substance 3D Designer is my go-to tool because I've spent countless hours mastering it, it feels like my comfort zone. I've created extensive material libraries from scratch with it, and it’s hard to quantify just how many materials I've developed over the years. What I particularly appreciate about 3D Designer is its flexibility, allowing for quick prototyping and testing of new solutions.

Material Techniques

For the Fabric Collection project, I employed three different techniques to create the materials: full procedural, procedural with input images, and photogrammetry.

Full Procedural: This technique involves using only Substance 3D Designer. I begin by gathering real-life references, which help me understand the fabrication process of the fabric. Understanding this process is crucial as it guides me in replicating the fabric digitally.

One key aspect is creating a detailed normal map effect that captures the fabric's fuzziness. This effect is achieved by blending a custom normal map effect in add sub mode with the regular normal map. This method effectively replicates the anisotropic light behavior on the fabric, giving it a realistic appearance. This technique is also applicable to other materials, like wood, where light behavior plays a significant role.

Procedural & Input Image: For this approach, the goal is to integrate patterns into the fabric in a way that makes them feel woven rather than simply printed. To achieve this, I duplicate the node used for the height map, which generates the normal map, and use it to create the tile sampler color. Ensuring the parameters are identical is crucial; otherwise, the strings won’t be colored correctly.

Once set up, I can plug an image input into the color map input, setting the color parametrization multiplier to 1. This allows the strings to inherit the image's color while maintaining the weaving effect. I also expose the image input, enabling me to change the fabric image in other software like Substance 3D Stager, depending on my needs. For the rest of the material, it follows the same process as a full procedural one.

Photogrammetry: I used photogrammetry for fabrics with complex patterns where recreating them procedurally would be too time-consuming. Photogrammetry offers a more efficient solution in such cases.

I used a cross-polarized process, which involves a camera with a polarized filter and a flash with a polarized filter at a 90-degree angle to eliminate glare and reflections during the scan. I chose RealityCapture for processing the images due to its ability to produce high-quality scans efficiently. Once the scan is processed, I bake it onto a plane directly in Substance 3D Designer.

Substance 3D Designer's baking tools are robust and flexible, allowing me to avoid transferring data between different software. From this base, I can apply procedural effects, such as the custom normal map effect I described earlier, or adjust color and normal balance to enhance the final material.

Each of these techniques was chosen to suit different aspects of the project, balancing between flexibility, efficiency, and the level of detail required. By combining these methods, I could achieve a comprehensive and versatile fabric collection that leverages the strengths of each approach.

Using Substance 3D Stager for the first time was a pleasant experience. I aimed to animate my renders rather than just take screenshots.

While the software is great for quick turntables, it has some limitations for more complex renders. I had to improvise with offset and zoom effects to achieve panning movements. Overall, it's perfect for fast, straightforward renders.

Conclusion

I was in my comfort zone with this project, so I think each fabric took me between a couple of hours and half a day to complete. The main challenge, which is the same every time I work on materials that will go together, is to ensure that the material works in association with the others.

It's simple to make a nice material, but making one that fits naturally with a bunch of others is another challenge. There is no magic solution for that, spending time comparing and adjusting your material is necessary to achieve the best result.

An advice I could give to artists who want to create great materials is that it's really important to have great references, preferably real-life ones because reality is always more complex and detailed than what our brain can imagine. More than just looking at it, it's important to understand how the material is made because that also gives some solutions on how to recreate the material in Substance 3D. This extra detail and complexity can make materials look much more realistic.

Elouan Harmand, Senior Material Artist

Interview conducted by Gloria Levine