jartz opened this issue on Mar 10, 2019 ยท 88 posts
bagginsbill posted Fri, 22 March 2019 at 7:08 AM
ironsoul posted at 7:59AM Fri, 22 March 2019 - #4348521
On reflection just boosting the spec value breaks the PBR model, as the upper surface of skin is a different optical properties than the lower maybe it cannot be done off just one node which means either layers or cycles node.
Sorry I'm not following this thought. If something needs two nodes, you don't immediately decide it has to be layers or cycles - we've been fresnel blending in FireFly for 12 years. What mattered in producing high-quality renders was to make the math do what physics does, and HOW (with one node or 100) doesn't matter. Obviously, if we can get the right math from one node, that is preferred.
The idea of using the physical root node is to produce a simple unbiased skin mat that is consitent across different lighting sets. The problem I found is the cross polarized image was much darker than expected (loss in the capture process ?) - typical the spec component should be around 4% but I boosted it much higher for above.
I am on the wrong computer and don't have the digital Emily resources at hand. What is the average value of the image in question? Was it around 10 out of 255? That is 4%, linear. Did you forget to tell Poser to use gamma of 1 on that map? Also, 4% is for smooth glass. Human skin surface is rough, and does not have an IoR of 1.5 but closer to 1.35 so we'd expect the specular map to tell us a reflection coefficient around 2.2 %. But do we even know what the specular map coding means? Suppose instead it has been normalized to represent the fraction of the idealized smooth surface fresnel reflection? In which case it would be encoded a lot higher than 4%.
It is unfortunate that PBR does not actually say what units and semantics are in a specular map - only thing we know is bigger numbers mean more. That's not telling us much.
Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)