Repin & Khaelen
Khaelen
Hey Repin, I've been mapping how light behaves in 3D space for rendering algorithms—ever thought about how the diffusion of light in oil pigment compares to a simple Lambertian model?
Repin
Your mapping is a fine start, but oil pigment behaves more like a layered glaze than a simple Lambertian surface. Each thin coat scatters light differently, so the overall reflection is a sum of multiple diffusions, not just one diffuse term. Try modelling it as a stack of semi‑transparent layers; that will bring your render closer to what painters of the 1700s achieved with oil on canvas.
Khaelen
Got it, I'll treat each coat as a thin film with its own albedo and thickness, apply Beer‑Lambert for absorption, and stack the BRDFs. That should give the layered diffusion you’re after.
Repin
Nice, but remember oil paint isn’t a perfect physical medium – the pigments scatter light in a way that pure Beer‑Lambert can’t capture. Try adding a slight anisotropic term for the glazes; that’s what painters like Caracci added by hand when they layered. And keep the thicknesses realistic – a millimeter is too much for a single coat.
Khaelen
Okay, I’ll add a minor anisotropic term to each semi‑transparent layer and trim the thickness down to a few microns per coat—no more than ten thousandth of a millimeter each. That should mimic the subtle directional scattering Caracci achieved.
Repin
That’s almost there, but Caracci would still not accept a 0.01 mm glaze – that’s too thick for oil. Aim for a few micrometers, and make sure each layer’s tint shifts subtly; the subtlety is what gives depth. Remember, the light should never escape the canvas like a screen – it should linger in the shadows.
Khaelen
Understood. I'll cap each layer at a few microns, shift the tint gradually, and clamp the light so it stays in shadow, not out the back of the canvas. This should give the lingering depth you described.