Trying to understand Index parameter in Gloss Material.

[gordonrobb]

I'm trying to work on a good setup for using files outputted from Quixel Suite. It gives me an albedo, goss, specular and normal. I'm happy with where to put them and even what to do with the gloss map to make it a good roughness map. My issue is the 'index' parameter.

The guide says that it is affecting the 'colour in the glossy input'. Firstly, there is no glossy input, so that confuses me. It looks for all the world like what it's doing most is putting a fresnel effect on the specular more than anything else. However, when I take a really good looking metal material from Quixel, and set up my Octane nodes, and then try to introduce an index of anything other than 1, it looks nothing like metal (unless I go way higher than it's supposed to be - referring to real world IOR values).

Can anyone give me any insite into the best way to use this parameter, in conjunction with a specular map which looks 'almost' right with an index of 1?

[fatrobotsneedlove]

Yeah, it does exactly what you thought. It determines the spread of the fresnel. Because there is currently no way to drive the IOR value with a map in Octane (something that will finally get implemented in V3 through OSL) it is impossible to use Quixel Suite/Substance Painter and get accurate results (well, without jumping through a ton of hoops exporting masks and setting up mix materials).

Now, can you make materials that aren't physically accurate but still look good? Yeah. Everything except metals have relatively similar IOR values. But if you have like a scratched painted metal? No bueno. Currently the best bandaid is to use a metalness workflow and use the metalness map to drive a mix material.

Why does Quixel keep saying they are working on an Octane export preset when it is currently completely impossible? I have no clue.

Oh, and FYI, if you do have a painted mesh with similar IOR values, you're still gonna have to tweak the roughness map to get it looking right, as no export preset in Painter or Suite currently matches Octane.

[gordonrobb]

I've got the roughness map sussed completely.

My issue is with the fact that the Manual describes what the Index does, and it doesn't make sense or seem to match. I can't even get a simple metallic looking ball, with now additional maps or scratches, to look right when I add the Index parameter into the mix.

[grimm]

You can simulate Fresnel reflections using the falloff node as Roeland demonstrates here:

viewtopic.php?f=21&t=33213

I have been able to bring in PBR metal textures from 3D-Coat that look pretty nice using his technique.

[gordonrobb]

Yeh, I'm completely at a loss about that whole post. I thought that Octane was physically accurate. Isn't the 'index' parameter what makes it physically accurate and therefore right? Why are we being directed to put in a cheat/workaround?

[grimm]

I don't know, but I suspect that this might be a placeholder for fresnel. But a dev could answer for sure. I do know that the devs have said that Octane does not yet support it.

[fatrobotsneedlove]

I'm confused about what you are confused about

The IOR sets the fresnel. Different materials have different fresnel values. In game engines, the fresnel is hard coded in. You can't adjust it (well, you can have two values I guess, as determined by the metalness map). In Octane you can, but you can only set one fresnel value per material because there is only a slider, no texture input. This makes Octane physically accurate in a way game engines are not.

If you applied an Octane material to your model and the Diffuse input only accepted one color instead of a map, then your object could only be one color. So similarly, because the IOR slot in the Octane material can only accept one IOR value via the slider, you can only have one fresnel value on your object. But metals and nonmetals have very different fresnel curves (well, every material has a unique fresnel value, but only metals are drastically different). So when you export a smart material out of Quixel Suite, there might be materials present in your map that have different IOR values. But because there is no texture input currently in the IOR slot in Octane, you cannot differentiate between these materials.

btw when you set it to 1 you turn the IOR off.

Are you confused about what fresnel is? (because it can be very confusing. ) would it help if I showed you how to make a metal ball like you mentioned?

[gordonrobb]

I'm not confused about what fresnel is. I meant I was confused as to why the octane team were recommending a cheat (in the thread you linked to). I know what fresnel is and what it does. I also know about different materials (paint and underlying metals) having different IORs. I can easily handle that will different materials mixed with masks that can also be exported.

My question is about how even a simple metal, with no other layers, can look quite good at index 1, but when you make it what it should be, it doesn't look remotely right. I'm trying to understand how to balacpnce the maps from Quixel, with the use of the Index. Forget about different materials. Just plain metal.

[fatrobotsneedlove]

The 'cheat' that grimm linked to is just a way to get certain metals with unusual properties looking correct. The example is gold, which actually has different fresnel curves for different wavelengths of light (if you look at the picture in the thread, you can see the effect on the edges of the tea pot on the right. They are slightly different color than the edges of the teapot in the middle).

The effect is pretty subtle. It sounds like the problem you're having is much more noticeable. Depending on how you're exporting your maps it might be an issue with your spec map. Best of luck.

[grimm]

fatrobotsneedlove is correct on that, Roelands method is for NK materials that have complex ior. I setup a test for this and the ior setting does control the fresnel reflections but it has a strange index range (1 to 8). If I remember correctly 1 turns the material into a mirror. IOR of 1 is good for grey metal but for colored metals is doesn't work very well, that is when you need to use Roeland's method. I use his method for grey metal as well as I think it looks better.

I set the diffuse to white (.85, .85, .85), the specular to 1, and roughness to 0.

Here is IOR of 1:

IOR 2:

IOR 4:

IOR 6:

IOR 8:

Roeland's method:

Hope this helps some.