[Fernando Tella]

Hi,

Is anybody using ior files lately? I have found some strange things with some files. When using, for example, most of the zinc files, which are supposed to be opaque materials, when the material is backlit it behaves like if it were translucent or had thin sss. With the more common ior files, like gold, copper and so it doesn't happen.

You can try in mxed: add a layer, change the preview to a backlit scene (leaf, for example), change to measured data mode, choose zinc_selenide.ior and refresh preview. Here it shows wrong.

Any one else noticed?

Thanks!

[eric nixon]

Confirmed here, noticed quite a few iors do it, also noticed those iors are VERY slow when applied to solid geometry? guess there are some buggy iors we never used/ noticed before..

Is a bit troubling

[Fernando Tella]

Thanks Eric. I noticed when using a zinc material that used to work quite nice in the past (hey where does that light under the roof come from? ) but it happens with plenty of IOR files. I wonder whats causing it.

[Half Life]

Germanium is particularly slow, I talked about this in one of my old tutorials as a reason using K value based metals was more practical than ior.

Anyway ior is fully capable of sss, which is part of the slowdown.

Best,
Jason.

[eric nixon]

Yeah its a bug in some ior files, at least HL confirms its not a new development.

BTW ior's dont have any sss.

[Mihai]

It could be because some of those IOR files where created by hdesbois by interpolating ND data and we included them with the install. Just a hunch but maybe that's what's happening...

[Q2]

Same happens to me with indium gallium antimonide(30 gallium).

That finally explains my lightleakage problems!

Anyone have a list of the affected IOR's which cause these problems so we can avoid them?

And is there a chance that NL will correct this in V3?

Cheers

Q!

[Fernando Tella]

I just bumped onto the same issue. A working ior list would be great to replace the current one.

[tom]

I was checking zinc selenide as a failing example and found decreasing 6.5 (3rd value in the header of the file) to 1.9 saved if from rendering transparent. We need to make sure if this is something to be fixed with these ior files or a bug of engine. Stay tuned.

[Fernando Tella]

I bet on bug. Those same ior files worked fine a long time ago, can't tell exactly when, with previous versions of Maxwell. I found the problem using a material from my library which I created and used without problems before (around year 2010).

[tom]

Makes sense, because that value 6.5 is technically cannot be wrong in any way. So, we're going to check what's wrong with the engine. Thanks for reporting it!

[tom]

Fernando, could you please check if you have this problem on closed volume objects? We think this is only problem with open surfaces...

[Fernando Tella]

No, it happens also on closed volumes. When seen from the outside it seems to be fine, but if you put a camera inside a sphere or a box, light leaks in.

Wrong trail.

[nachob]

Hi Fernando,

Its similar to the behavior you see when applying a dielectric or a SSS to a plane, it may look similar but its no correct. In the way it works here, the IOR could have dielectric properties and the engine is acting as the object where its applied has thickness. In your example, putting you camera inside a cube is like putting inside several planes, so its the same problem. For an IOR to work correctly must be applied to objects with thickness.

nachob

[seghier]

maxwell can't render zinc sulfide (cubic)
i open it with text editor and find values separated by commas; i remove the commas and worked but give result like sss material