The color of heated metal isn't arbitrary - it follows physics. Here's the logic behind it and why it matters for material work.
Note: This started out as a 'quick' post but I went down the rabbit-hole trying to be as accurate and precise with my findings.
The Science
When steel is heated, a chemical reaction occurs between iron and oxygen, forming a thin oxide layer on the surface. The thickness of that layer - determined by temperature and how long the metal is exposed to heat - dictates the resulting color. Thicker layer, different color. It follows the same thin-film interference principle behind iridescence on soap bubbles or beetle shells.
The key insight for material artists: because temperature controls the layer thickness, the color progression is always consistent. Hotter areas produce predictable colors, cooler areas produce others. The gradient isn't artistic - it's physical.
The Color Gradient
Scientifically rigorous resources on heat coloring are sparse, but this reference by Jeffrey H. Dean is the strongest visual breakdown I could find. It's grounded in an artistic practice where craftspeople control temperature precisely to achieve a target color - which means the color-to-temperature relationship is treated seriously, even if the context is craft rather than physics.
The color range follows a consistent progression - pale yellow → straw → gold → brown → purple → blue → grey - with each step corresponding to a thicker oxide layer and higher temperature. The gradient always runs from the hottest point outward. That directionality is the rule to encode in your material.
⚠️ This progression is specific to steel. Aluminum, copper, and gold each produce different color ranges under heat - the underlying physics is the same, but the output colors differ.
That color shift also has consequences for how the surface reflects light.
Effect on Reflectance/Specular
The oxide layer affects reflectance in a way that's worth understanding clearly: based on my research, it behaves like a thin-film coating, producing the color shift through interference rather than by changing the underlying material. The surface remains metallic - the oxide layer doesn't convert it to a dielectric. Metalness stays at 1. The change is in the albedo and a subtle shift in the specular response as the layer thickens.
Color & Temperature Reference
Consolidated from Machinery's Handbook (5th–20th editions), the ASM Heat Treater's Guide, Evans (1925), and Bhadeshia's Cambridge notes.
Replicating in Substance Designer
The gradient logic translates cleanly into a Substance node or filter. The interesting part is that you can drive it with physically based temperature values directly - expose a temperature parameter, map it to the color gradient, and let that control the output.
For the spatial variation, the inverse of a thickness map generated of your asset, works well as the input. Thinner areas of the mesh heat faster and reach higher temperatures first - which means they sit further along the color progression. Feed that into the gradient and the directionality takes care of itself. The same rule that applies in the real world applies in the node graph: thickness controls temperature, temperature controls color.
I normalized the color range so it could be easily remapped to a gradient. By using the aforementioned inverse of the thickness map, I can dynamically control the positioning. Wrapping this in a custom node would add more controls - a min/max remap of the input mask, for example - but the core logic holds as is.
© 2026 Stefan Groenewoud - All views are my own, not those of my employer.