Shading bug behind glass?

[roeland]

The problem I think is that alpha shadows is disabled in the kernel settings. This switches off some calculations needed for shadow calculations through transparent objects, either with an alpha channel or glass with fake shadows enabled.

--
Roeland

[treddie]

I had already tried that. So I went back and tried it again to see if I actually missed that, but same problem.

[treddie]

Any ideas? It's been about 3 weeks and I have a rendering I can't complete due to this problem.

Thanks!

[roeland]

I checked your scene again, and if you enable alpha shadows in the kernel settings, and fake shadows in the glass material, it renders correctly.

--
Roeland

[treddie]

OK, I see that that works. But I am confused as to what is going on. I did what you said and enabled both the kernel's Alpha Shadows and the material's Fake Shadows options. It took both to get it to work.

But the manual states that Alpha Shadows is only used with Alpha maps which I am not using:

"If alpha maps are used in the scene, this setting controls whether the shadows will be calculated from the mesh geometry or from the alpha map."

And it also states that Fake Shadows is only used with architectural glass which I have assumed was a faster less accurate glass material:

"Fake Shadows is a Boolean value that sets the architectural glass option for all meshes sharing that material."

There is only a brief description in the old pdf manual about what architectural glass does (description completely absent in the online manual), and why the "regular" specular material (archi glass off) does not quite work in all instances:

"When enabled, the specular material exhibits the characteristics of Architectural glass with its transparent feature allowing light to illuminate enclosed spaces..."

The regular specular material already lets the transmitted light illuminate an enclosed space, since it is a transparent refractive material. Why is architectural glass needed by switching on both Alpha Shadows and Fake Shadows? Why are the shadows "fake" and I assume, no longer unbiased?

[treddie]

Can anyone at Otoy chime in on these questions? The manual is absent on this.

[treddie]

OK...The problem is not as easy as the above would suggest. In my initial test, I found that the recommendations SEEMED to solve the problem. But when I let it render longer, the problem was still there, just not nearly as bad. Finally, although switching Fake Shadows and Alpha Shadows on seemed to increase the amount of light passing through the specular material, I found that the problem had less to do with that, than with the PMC settings. See attached. By dropping Exploration Strength and Direct Light Importance way down, this had the greatest affect on the problem. Yes, turning Fake Shadows/Alpha Shadows on dramatically lightened up what was underneath the glass, but the rounded edges became more natural looking due to the PMC changes.

BUT, what happened to refraction?! I think this is a totally unrelated problem. If you look closely at either image, there is no offset image of the underlying surfaces seen through the glass. It is as if IOR is set to "1". In both images, the Index is set to 1.45, plenty to create the expected offset. Look along the glass edge and there is no refraction bending at all. I don't think this is an Octane problem, but I'll be damned if I can figure out what it is. No matter if Fake Shadows/Alpha Shadows is on or not, refraction is zero. Also, I have checked the surface normals and they are fine...pointing away from the glass surface on both sides. What the heck?!