day and night test

Hi guys, I did this test last days and I get an unexpected results with refraction in the glass fences …


Can you mark that area where “unexpected results” are visible?

The material has 1.54 Ior for reflection&refraction but as you can see, the displacement its to big.

Could it be that your glass fence has no thickness, yet material has refraction enabled?

Yes, it has 1cm of thickness. It is a extruded spline with shell modif. I allready check the normals of the polys and all its ok.
And the material is a standard glass, with ref/refrac. I didn¡t use thin glass.

Can you save the problematic part of the scene (just the objects+materials that are acting strange) or recreate it in some new scene and upload?

One more guess: does your glass material is part of blend material?

No, its normal material without blend/layers/maps

Sure, tell me where

It can be here if you are fine with it. When writing post, click the “attachments and other options” text. Making the scene as small as possible would be appreciated. :smiley:

It is
unexpected_refraction.rar (609 KB)

hmmm ..is it a CAD imported geometry? ..

i was looking at your file and you extruded isolated edges, I cannot see any splines as you said. Doing this gives a lot of problems with materials, I dont know why, but here in the office we often have this sort of problems importing geometry from revit or autocad. The final geometry tend to get bad xform, bad uvmapping, etc, etc …so, try to import in a different way so you can keep the curves from cad as a spline in 3dsmax …

or, you could extract the edges with the “create shape” button so it can be turned into a proper spline…then you should use a “sweep” modifier to make the glass fence, this way you can uvmap the glass with some noise using exact measures like mm …

----------wait …i was looking it again and your fence railing is a proper spline…you should use that with a sweep modifier, its a lot easier.
and your “glass boxes” geometry should had the “THIN” ticket in the material, and probably clean that mesh, because those are boxes within boxes…I don’t thinks that’s useful for its purpose here

Are you sure? In file you attached all glass materials are nested into blend materials. You can see in ME previews that it leads to problems.

Either tick “thin glass” option or remove glass materials from blend.

I didn’t saw this. sorry. Now I solve the problem.
I’ll put the final image here.
Any coment is apreciate.

I´ll try to solve this too.

Thanks for your coments

I checked the scene too and didn’t see any problems with geometry (I think boxes inside boxes should work fine - same for bubbles, right?) but fixing refractive blend materials should help.

I solve the my blend error. I create a new simple material and It works perfectly (thaks for this).
Now Im testing the scene with “thin glass” , wi’ll see

I’ve learned not to trust imported geometry. They work well in some cases but not all that’s imported its usefull.
For instance some glass boxes clash with the glass fence, so its far easy to edit just one box (i’ll do extruded close splines probably) …

I render a “thin” scene and “no_thin” scene, and here is the result:

Both scenes are 99 passes, with “thin” glass take 16"36’ and with “no thin” 27"00’.

The direct_light pass look’s good in both, but the indirect looks a bit dirt. what can I do to solve this?

And the shadow_pass has the same. If I increase the light subdiv. can I solve this?

Thanks in advance for your help


Direct pass is dependent on light samples multiplier.
Indirect pass - on GI vs. AA balance.

So if you want to improve the quality of indirect lighting within given amount of passes - you should increase GI vs. AA balance.

You can also try decreasing Light Samples Multiplier (but not much, I would say not less than 1) and see if the quality of direct lighting is visibly worse. It should make the rendering slightly faster.

Thanks Maru, I’d try with 32 GI vs. AA and 10 Light Samples Multiplier