Hi,
I just bumped to a video explaining the somewhat misnomer design of the Principled BSDF Shader.
Just wondering if, do you use the FresnelToSpec Node (IOR Fix)?
Regards,
Ben
No. Having it in a 0-1 range makes it much easier to manipulate. The effect on reflection is not what’s going to make or break my results. It’s not a misnomer design, and I agree with the design choice. I’ll also happily use facing->pow 5->add 0.05 for architectural glass in order to not having to deal with 1/IOR values for backfacing faces.
2 Likes
Ah gotcha. So I guess that FresnelToSpec Node is only for specific cases (i.e. archi glass) and not as a general fix.
I haven’t tested it on archi glass as I don’t use it, so I wouldn’t know - I fake the fresnel curve of glass using facing->pow ->add 0.05 instead for a curve very similar to IOR 1.55 (glass’ish). I don’t like the idea that it somehow makes things better, or more “accurate”. Since I frequently modify specular to match what I see, having a value in the 0-1 range is way more easy to manipulate as the maths can be previewed. IOR is “too theoretical” as the microstructure causing specular energy loss (in regular materials) can’t be simulated by simply disturbing the flat surface, but it’s easy enough to model such an example where no amount of roughness will cause enough specular energy loss.
One thing to keep in mind is that any time you use fresnel node or layer weight/fresnel, is that the backface will use 1/IOR. A single glass pane with the wrong normals will let you see snells window (water surface from below), and shadow calculations (sun shining through window) will also have this problem. Faking it solves all these problems.
2 Likes
Ah gotcha.
I didn’t know about this one:
that the backface will use 1/IOR
Thanks for the heads up.