Creating automotive CMF materials in 3D is less about reproducing a color and more about reproducing how a surface behaves. A black textile, black leather, and black injection-molded polymer can have very similar base colors, but under light they should look completely different.
Their roughness, grain, weave, reflectivity, and surface scale are what tell us which material we're looking at. That's why a useful automotive material generally needs more than a single texture. The problem is that manually building all of those properties takes time, especially when we're still exploring. With twirl, we can describe the surface we want and generate a complete seamless PBR material first, then spend our time refining the directions that actually work on the vehicle.
Start by describing the physical surface
Before creating anything, it helps to describe the material the same way we'd describe it to another designer. For an automotive polymer, that might include the grain, grain scale, color, gloss level, and surface uniformity. For a textile, we might care more about weave direction, fiber size, and color variation. Leather introduces pores, natural grain, wrinkles, and differences in sheen.
Those characteristics translate naturally into a PBR material. Base color handles much of the visible color information. Roughness describes how concentrated or spread out reflections are. Normal and height add smaller surface structure, while metallic controls whether the material behaves as a metallic or dielectric surface.
Autodesk's implementation of Substance materials in VRED supports this type of metal/roughness workflow, including base color, roughness, metallic, normal, and height information. Autodesk's Substance material documentation is a useful reference if VRED is part of your visualization pipeline.
Turn that description directly into a material
This is where we can skip a lot of repetitive setup. Instead of searching for a car plastic texture, downloading something close, and then trying to turn it into the surface we actually wanted, we can start with the description itself.
twirl turns that description into a seamless PBR material with the maps we need to start rendering it. We can apply it to the dashboard, door panel, or center console and judge it in context. If the grain is too fine, we can generate another direction. If the surface feels too matte, we can describe a satin version. We don't have to treat the first material as something precious that needs to be endlessly adjusted.
Scale matters more than detail
One of the easiest ways to make an automotive material look wrong is to get its scale wrong. A dashboard grain is usually extremely fine. A woven seat textile might have much larger visible structure. Brushed aluminum has directional marks that need to make sense relative to the component.
An otherwise convincing texture can look completely wrong when its grain is several times too large. This is why we like getting materials onto the actual model early. Material spheres are useful for checking the maps, but a material intended for an IP should eventually be evaluated on an IP.
Roughness usually does more work than we expect
When a material feels wrong, it's easy to start adjusting the color. For many automotive surfaces, roughness is just as important. Two dark plastics can have almost identical base colors but produce very different reflections, which is often enough to make one feel softer, cheaper, more technical, or more premium than the other.
The same applies to leather, fabric, and painted surfaces. Roughness variation helps break up reflections and keeps a material from looking perfectly uniform, but the variation should still be tied to the surface we're trying to represent. Random noise gives us variation, but not necessarily a believable material.
Keep normal maps subtle
Normal maps are another place where automotive materials can become exaggerated quickly. A fine injection-molded grain should affect the way light travels across the dashboard without making the surface look carved. Leather pores should be visible at an appropriate viewing distance, but they shouldn't become large craters when the camera pulls back.
A good test is to move a broad light across the surface and watch the reflection. We should see the material structure influencing the highlight without every individual feature demanding attention.
Generate variations instead of perfecting the first result
This is where generation changes the workflow more than simply making one material faster. We can create several related directions and compare them. Maybe the same charcoal polymer gets a fine organic grain, a slightly larger geometric grain, and three different roughness directions. Maybe we generate the same woven structure with warmer and cooler fiber combinations.
In the ArtCenter and Rivian transportation-design workshop, students explored variations in pattern, gloss, translucency, and layering as part of their digital material workflow. Adobe's workshop recap shows the value of seeing several plausible material directions in context rather than overdeveloping the first one.
That's the kind of iteration we want twirl to make easy. Instead of spending most of our time creating options, we can spend more of it comparing them.
Move toward accuracy when the design needs it
Once we've found the material direction, the requirements change. If we're reproducing a real supplier material, we may need measured appearance data, scans, or a more sophisticated authored material. VRED supports measured AxF and BTF materials, MaterialX, MDL, and other workflows for this reason. Autodesk's material overview covers the broader set of options.
twirl is useful before that point. Generate the ideas, test them on the design, and find the few directions worth developing properly. Instead of making every possible material accurate, we can spend that effort on the materials we've already decided are worth keeping.