Author Topic: Spec/Gloss to Metal/Rough  (Read 15261 times)

I am trying to convert a spec/gloss material to the metal/roughness model, as I want to convert a Megascan from Quixel to a substance to be used in painter.
So far I was able to find the opposite node only (MR -> SG), is SG->MR somewhere?


The other way is a bit more complicated, we still have to work on it.

Product Manager - Allegorithmic

We have not included such a converter yet because it's not as simple to convert in that direction. The "Spec/Gloss" shaders cover a wider range of materials (even including some that "don't exist in reality"):
If your "Spec/Gloss" shader does not enforce energy conservation (for the total specular+diffuse reflected power), then there are combinations of textures that have no equivalent with "Metalness/Roughness" (e.g. full white diffuse and specular).  And even if it does, what should we do with materials with low-valued but colored specular which "Metalness/Roughness" can't represent (because non metals don't have colored reflection) ?

Thank you for your explanation, I didn't know about it.
Honestly I am not very fond with PBR materials using the Metal/Roughness model, as I am used to the Spec/Gloss one. Time to study the Disney model i suppose :)

I suppose that for the same reason would be difficult to support the Spec/Gloss model (in the brush properties) inside of Substance Painter.

The "metalness" model isn't that complicated (I find it simpler). You just have to decide whether the surface you are seeing is a metal or not, and whether it should show sharp reflections or not. It leaves less room for human error.

Hi Cyrille, I gave the Disney model a lecture.
So, from what I understand, dielectric materials have b/n specular, with a fixed value to 0.04, and metal have black diffuse with colored specular (or in other words, the specular is stored in the albedo).
So I can see why it could be a problem to convert from this one to the spec/gloss model.
I also have a question to see if I understood correctly: I found that a rust material in this chart has a colored specular. But usually the rust is considered to be dielectric, and without that colored specular, in the metalness model there is a loss of saturation.
Am I correct or I am missing something important?
Last Edit: June 24, 2014, 10:05:39 am

You are correct.
Pure rust (ferric oxyde) is a dielectric. I suppose that in the samples scanned here each texel cover a small surface with both rust and a bit of iron showing, which would explain that you have a slight colored specular 'on average'. You can do the same with the "MR" model by using a "metalness" value which is not pure black or white but the ratio iron/rust in the texel.
The same reasoning applies if you have dust on a metal, or very small amounts of dirt, fingerprints, etc...

Thank you Cyrille!
I really appreciated your help :)

Ok if I take the front facing value In the specular. I can get my ior with this.

If I got my IOR I can determine my metalness map ??


Something like this.

I keep a little grading with this approach!