Reflection Glossiness Weirdness

[Steffen Dünner]

Hi guys!

I just tried the different BRDFs implemented in the V-Ray Material in MODO.
I expected to see a difference, but is this supposed to look like this?

All cylinders have the same glossiness of 0.8. BRDFs from left to right are: Phong, Blinn, Ward, VRay, GGX



Here's a glossiness of 0.95. The left cylinders are almost perfectly glossy, while especially the one with the VRay BRDF is still very rough.



To get a glossy result with VRay or GGX you're basically in the range of 0.98 - 1.0 which is kind of crazy if you're thinking of the range of a usual glossiness texture, or am I doing / understanding something wrong?
I had no chance to test the same with V-Ray for Softimage, but I'd expect a glossiness of 0.5 to look "half glossy" and 1.0 to be completely diffuse.

Cheers
Steffen

[talleman]

That looks pretty accurate. I remember reading somewhere that the vray brdf was experimental and needed to be removed.

The blinn and phong are acting as you would expect, while the ward is usually used for metal materials (but still correct).

This gives a pretty decent explanation of ggx http://www.shlyaev.com/rnd/37-cpp-category/54-ggx

[Steffen Dünner]

Thanks!

Over all I was wondering about the value range vs. the visual results I'm getting. Doesn't feel very artist friendly at the moment, especially for the Ward and GGX BRDFs. The "usable" range of values for these BRDFs lies somewhere between 0.5 - 1.0. Everything below 0.4 looks basically completely diffuse. Other renderers I'm using (Arnold, MODO, Redshift) give me "half glossy" at 0.5, while GGX at 0.5 produces the result I'd normally expect at 0.1.

This e.g. is MODO renderer at 50% roughness (=0.5 glossiness):


This is V-Ray, all possible BRDFs at 0.5 glossiness, which is pretty rough already:


In document you linked, just take a look at these images:





I posted an image with 0.95 as glossiness value and it's nowhere near the glossiness in these "official" images. Apart from that, the Blinn BRDF in my images reflects sharper at the same glossiness value than the GGX BRDF, while in the second official image it looks sharper than Blinn (which is more like what I expect as well).

I also checked the vrscene coming from my MODO scene and the values sent to V-Ray are the same as in the GUI, so it's not some sort of "translation problem".

To make a long post short: The results from the different BRDFs are perfectly OK, it's the value range that feels strange. I just wanted to know if that's "how it is" or if there's a problem under the hood.

Cheers
Steffen

[vlado]

What you got in Modo is a first draft of the GGX BRDF. For subsequent builds, the glossiness is remapped differently to better match the other BRDF types.

Ignore the V-Ray BRDF though, as mentioned above it is just an experiment.

For your comment about other renderers, they artificially clamp the possible range of the BRDFs. Sometimes you do need to get rougher materials and a common complaint has been that in other renderers, this is not possible.

Best regards,
Vlado

[Steffen Dünner]

OK, I can perfectly live with that. Thanks for the explanation. Remapping the input range sounds reasonable to get consistent results while switching BDRFs in V-Ray.

Thanks
Steffen

[vladimir.nedev]

The new GGX glossiness formula is now part of the V-Ray for MODO branch, but our nightly build machine failed and I
will be on vacation next week, so it will be at least a week and something until you can try it. Here is a comparison
between the old and new formula.

Old :

New :


The new formula also has a "tail falloff" parameter that let's you control the glossiness
remapping even more. Here the tail falloff is set to 4.0 :



Greetings,
Vladimir Nedev

[Steffen Dünner]

Wow! This looks very promising!

Thanks for implementing this and enjoy your vacation!

[Steffen Dünner]

Cool! I just installed the latest nightly (3.00.03) and glossiness with GGX is now a pleasure to use.
Thanks

[la_de_longe]

Where download nightly versions?