I have a scene that has a piece of foil in it, from a cigarette packet. It’s a simple piece of geo with displacement mapping and bump mapping. I have the lighting and shading looking how I want it at 1080, but when I render at 4K it looks very different. At HD I get nice hot white highlights/reflections of the type you’d expect to see on such a material. The foil looks bright and of the correct lightweight paper-backed type.
But when I go to a 4K render, the appearance of this material changes drastically. The hot highlights are lost and it renders looking more like heavy foil, grayer and more like cooking foil that has been compressed.
Lighting is simple, using a dome light with HDRI mapping and a single rect light.
I know that resolution has always affected shading results to some extent, however this seems to be a very extreme case. Is there anything I can do to mitigate this, other than reworking the shader at 4K to try to get it back to where I want it?
In most cases it might just result in a slightly different *looking* render as a result of more pixel information in the image, e.g. more details from a file texture being sampled etc. However, this seems to be changing the reflective qualities of the material completely.
I think dome light hdr can’t be used in reflection if you are after some high res reflection results since dome light hdr gets compressed a lot when a blurry reflection is involved. So for that reason, we usually use dome light hdr as lighting, and override reflection with full hdr in the global environment. Not sure how this would affect your case but worth a shot imho.
I’ll give that a shot -- I am reflecting the HDRI through the light currently but have also worked the way you describe sometimes. In the meantime, I’m able to post this image to show the problem.
I’m guessing it has to do with the displacement map. Try disabling it for testing. Higher res would give a different result as there’s more displacement details to fill in.
OK, I think I’ve traced this to bump mapping. I’m using Maya’s procedural bulge to create the overall grid texture, on top of displacement. For some reason when I go to a higher resolution, the bump value on this needs to be cranked way down. Bug?
Just a guess, but maybe when the resolution is higher, the bump map is sampled more accurately, and what you see is closer to a true representation of what your material actually looks like. When the resolution is lower, the tiny details in the bump map are mushed together, and the renderer sees a more uniform/smooth surface, which means the large scale geometry has more influence over the reflections.
If that’s the case, you could try scaling the bump map up and lowering the strength, so it becomes more uniform at higher resolutions as well. Or you could bake the current bump map as a texture map, which would do something similar by limiting how detailed the bump map can be no matter how closely it’s sampled. I think either option would help stabilize the look of the shader.
Yes, I think you could be right that it’s the procedural causing the problem. I’ve had strange results with procedurals and bump in the past when uprezzing now that I think about it -- it was the apparent change in reflectivity that was confusing me this time. I’ll try baking it to a texture.
It’s not as much reflectivity as it’s glossiness, I think you accidentally created a very diffuse material with lots of tiny detail in your bump
I’m curious to see if the texture baking method works, do keep us posted.
Not if the displacement settings are set to pixel (ie. screen space.)
V-Ray will try and hit the value (say, 4px edge length), whatever the resolution, meaning that higher resolutions will generate more triangles, and potentially change the shape quite a bit, especially if procedurals, rather than bitmaps (for which V-Ray compensates quite well), are used.
If you switch to a world-space approach, then the displacement will remain consistent across varying resolutions, with all that that entails.
Note: this isn’t in reply to your original issue, i just wanted to make this bit clear.
I saw another post related to this issue somewhere on here, and someone suggested that the Bump Delta Scale comes into play for this problem. I haven’t tested it and this project is now finished, but perhaps it might have helped.
