VRay Glass Facade - Splotchy Reflections/Refractions

#1

Hey there! Hoping that I’m posting this in the right section…
I have a question about rendering reflective/refractive surfaces with VRay in 3ds max.

I study architecture and I needed some exterior visualisations for my project presentation. I was making some test renderings after I’ve set the materials (used only VRay materials) and lighting (VRay HDRI env. map, Daylight with VRay Sun), when the problem occured:


I created a double facade system with aluminum and frosted glass panels above a concrete structure. The facade is a group of editable polys with 10 cm offset from the structure. Each polygon is assigned with a matID (1-4) and there’s a sub-object mat applied to it. The glass mat is a simple glass with fresnel and 0.9 reflection/refraction glossiness added to achieve the frost effect. The surface behind has a concrete tiles mat assigned to it (some reflection, bump and diffuse maps…).

You’ll notice the splotches on what was supposed to be a frosted glass surface. It’s interesting that the problem seems to disappear when the distance between the glass and the object behind it increases (notice the staircase, which seems to be correctly rendered).

I tried increasing the subdivs in materials, tweaking primary (Irr.Map) and secondary (Light Cache) engines, antialiasing, reflection/refraction level and some glass mat parameters (affect shadow, reflective/refractive amount…). As nothing worked so far, I’m guessing that it’s either a VRay material issue or there’s not enough light info for VRay to correctly calculate the reflections of the two nearing reflective surfaces (even though I increased the max depth level to 18) … but then again, it could be my error. I’m using 3ds max 9 and VRay 1.5.

Looking forward to your comments!

#2

Are you using interpolation for glossy effects? And what’s your glossy amount for reflection/refraction set to?

#3

Nope. No interpolation used. Here’s what else is confusing … when I changed the reflection/refraction glossiness values to 1, this rendered out:

Here are the glass params:

I also tried changing the environment (different HDR, Sun position) and some materials (replaced concrete with a more simple mat - without panels, no reflection), excluded the falloff map on the glass… and still no result

#4

Looks like sampling issues,

Does the glass have thickness?

First thing I’d do is ditch the HDRI (I have a feeling that’s whats causing it) and use vray sun / with dome light.

Frosted glass material : I’d start over -put diffused to black and Refrac/Reflect to white.
leave everything default, check the Fresnel reflections box.

put the refraction glossy to about .85 (adjust to desired look)

you don’t need to go crazy with subdivisions, just going to slow your render times down with out gaining, 8 subdivisions should be fine for glossy around .8

also if your not using SP3 defiantly use:
http://www.scriptspot.com/3ds-max/plugins/autogamma
For linear workflow

#5

Hey, thanx for replying!

  1. Thickness: No, the glass doesn’t have thickness - tried adding thickness (3 cm) and re-rendering - no result

  2. Material: tried reseting the mat and configuring it as you said, but the result is pretty much the same

  3. HDRI - tried replacing the environment with VRaySky - no result. However, even if it did produce results, I’d probably have to find a workaround, since I use the HDRI as my main color - light - reflection - refraction - GI solution and the Sun only for adding some extra light (HDRI only images seem to render out a little bit underlit). In addition, somehow it doesn’t seem as an environment map problem, since the splotchiness seems to disappear when the distance between the glass and the object behind it increases.

  4. Autogamma - WOW! thank you very very much, I’ve been searching for some quick LWF script for quite a while now, this helps a lot!

Anyways, thanks again, hope to solve this thing soon =D (deadlines and so on … )

P.S. I’ve narrowed it down to refraction - after rendering quite a few different versions of the glass material, I concluded that the reflections were more or less okay, while the refractions seem to cause problems, and, in addition, only when the object behind the glass is in the proximity to the glass itself. Also, I noticed that the orientation nor the sun position don’t matter, since the problem occurs on all sides of the model. Hope this helps.

#6

Does it happens with irradiance turned off? If the problem is coming from the irradiance, try using the option “check sample visibility”, this can be an interpolation problem.

#7

How close is the glass and the surface behind it. There are issues with rendering coplanar faces. Try raising the Secondary Ray Bias by a small amount like 0.001-0.002. If that does not work, try giving the glass more thickness so it overlaps the material behind it, rather than having the material barely touch the one behind it.

#8

Thanks for the replies again!

On the plus side, all three of the suggestions do fix the issue a bit. On the other side, however, neither one removes it completely. It seems like we're on the right track =D

Firstly, I adjusted the sec. ray bias to 0.002. This reduced the effect a bit but introduced problems with other reflecting surfaces (road behind, floor tiles), causing them to reflect rather strangely and produce overburnt surface parts.

Secondly, I set the GI engines to QMC for primary and LC for secondary. This somewhat reduced the splotchy effect, but, only in parts farther away from the camera's position. Result is pretty much the same on surfaces closer to it. In addition, with QMC being my primary GI, it also introduced some overall grain, but I believe this could be dealt with by increasing the subdivs.

Thirdly, I reverted the GI engines to Irr. Map (Medium, 50/20) and LC (1200, 0.01) and checked the check sample visibility checkbox. This also reduced the overall splotchy effect, but not completely. Then I checked the show samples checkbox and this rendered out:

[img]http://img246.imageshack.us/img246/7484/splotchysamples.jpg[/img]

What I find interesting here is that everything seems to be in order (?!) while the rendered result, obviously, doesn't. 

In addition, I believe that we could actually split the issue in two parts, possibly not with the same underlying cause: If you look closely at the picture with non-frosted glass, you can notice two things which seem to be incorrect:
1. black splotches around the glass border
2. white triangle-like artifacts randomly positioned across the glass surface
Decreasing the refraction glossiness value seems only to dilute the surface effect, while emphasizing the one on the border.

P.S. @ rmejia:
I’m not sure what do you mean with increasing thickness and surface overlapping =D
The facade doesn’t have thickness applied to it - it’s only a grid of single polygons, with their materials controlled by the sub-object mat (with 4 Vray-sub-mats assigned to it). The geometry is offseted from the structure behind it by 10 cm. Structure walls have a simple concrete/stucco mat assigned to it (noise bump, 20 reflection, 0,85 glossiness… similar to the mat you see on the benches).

#9

I’m not sure if I understood correctly the original post, but I was under the impression that the glass was slightly in front of the facade. That behind the glass is the facade, not a hole in the wall. If there is no hole behind, then the glass should be thick and overlap the facade wall, as in the attached illustration, or it should be farther apart from the wall.

If the geometry is like in the third diagram, then it should be rendering ok.

Edit: Attached a max rendering, with 4 glass planes. The first one is flush, the second one slightly separated, the third one more separated the fourth one has a shell modifier. It does not look like yours, but you can see what happens when the two planes are close together (second from left to right). Which is what I thought was your problem.

#10

Wow, thanks a lot! This actually solves the problem! =D

However, I do have one more question. This is an approximation of the current and proposed geometry (although in the rendering you can notice the glass panels in front of the structure wall - which seem to be causing problems, but I don’t want to change this for non-cg-related reasons):

I really don’t understand how come this is the solution =/ I mean, it does solve the problem … but, on the other hand, it’s entirely physically incorrect, if you think of it. What’s bugging me is that the first solution, even while being physically (more) correct, still doesn’t render right. Does VRay have problems rendering coplanar faces in all cases or do I have to change my workflow 'cause I’m doing sth wrong?

I’d be really glad if someone could explain what’s at hand…

Anyways, thanks for the effort, you’ve helped a lot!

#11

V-Ray and other raytracers do have problems rendering coplanar faces. There will always be problems when there are two overlapping geometries (a copy of a plane exactly over another) or with 2 geometries very close together as in your case. I am not exactly sure what the cause of the problem is, could be the light ray bounces go crazy with 2 surfaces that close together. If you hace access to the vray forums you could get a lot of help over there: http://www.chaosgroup.com/forums/vbulletin/

The diagram labeled suggestion is what I would do. I think you can apply a shell modifier to all of the glass at once to quickly give all of the glass in the facade thickness, extending it into the wall. It is not physically correct, but it will render correctly and temporarily solve the problem.

Looking back at your post’s, is the material you posted with the VRayEdges and Falloff map the glass? Have you tried a simpler glass material to see if it works?

#12

Yeah, I completely agree - that is obviously the case. I know that forum, will look into it :smiley:

That is exactly what I did (PolySelect -> polygon -> byMatID (1) -> Shell (inner amount 20, straighten corners)) - and it rendered correctly, thanks a lot!

Originally, yes. However, I tried replacing it with what HeadSmell suggested (post #4) and it rendered pretty much the same. Then I amused myself with tweaking mat’s parameters for quite a while, changing the reflection/refraction amount, glossines, removing the falloff map, removing fresnel … and I concluded that the refraction is most likely to be causing problems, since in pretty much all renderings, reflections rendered out correctly. I even thought that perhaps I accidentally reversed the normals, but that was not the case. It was obviusly not my fault entirely - some of it has to go to ChaosGroup :smiley: :smiley: :smiley: just kidding

All in all, problem (at least temporarily) - solved! Again thanks a lot, you’ve been much helpful, hope that I could share some useful advice on this forum someday… :smiley:

#13

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.