Page 1 of 2

help: odd pattern on glass plate

Posted: Wed Jan 31, 2007 4:30 pm
by -Adrian
I'm getting some kind of pattern on a plain box of glass that shouldn't be there:
Image

I made sure the mesh doesn't intersect with others and switched between several different environment maps, no luck.

Any ideas welcome.

Posted: Wed Jan 31, 2007 5:33 pm
by joshh
Newton's rings ?
maybe Maxwell is too physically accurate?

Posted: Wed Jan 31, 2007 11:53 pm
by -Adrian
That could well be it, i haven't witnessed it ever before. The glass plate mesh is perfectly straight and doesn't even contain edges that would allow a bent shape (a necessity for the effect i read).

Using AGS solves it but looks really cheap.

Here a full render showing the effect nicely:
Image

Arrgh Maxwell, i don't know if i should be bothered or amazed :P

Posted: Thu Feb 01, 2007 2:10 am
by joshh
I've seen this effect in photography when two flat pieces of glass are touching, but as you said, there's nothing touching. How was your original glass material made - is there a coating?
Very nice object design and modeling btw.

Posted: Thu Feb 01, 2007 1:09 pm
by Tim Ellis
Very strange. :shock:

Any chance you could post an mxs with just the glass section & lighting?

I'd be really interested to disect this.

The really strange thing for me, is the edge of the glass nearest the camera, halfway up it looks as though a normal is inverted, but it's one of the concentric rings that has been cut in two. :?

Cheers,

Tim.

Posted: Thu Feb 01, 2007 1:32 pm
by Fernando Tella
Does it show from any other angle?

Posted: Thu Feb 01, 2007 2:36 pm
by tom
What's the Nd and what happens when you change it?

Posted: Thu Feb 01, 2007 3:22 pm
by -Adrian
Does it show from any other angle?
It seems to disappear when the angle gets less steep, but not right away, maybe at 30° less or so which changes the whole idea i had for this render.

When moving the glass out far enough (away from the rest), it's pretty much gone at steep angles at well, but still the glass doesn't look normal imho. Very strong ring reflections remain at the short side of the glass, that is closest to the camera.

The glass has a microbevel, i hope that isn't the cause.
What's the Nd and what happens when you change it?
1.51 (Wizard default)

The phase of the pattern changes so to speak, but it doesn't vanish at all.



I'm gonna experiment a bit in Modo and see if there's some mesh error.

Posted: Thu Feb 01, 2007 5:57 pm
by lllab
cool!
if you want to get this you would never get it;-)
cheers
stefan

p.s. are you sure there is nothing doubled, faces or polygons?

Posted: Thu Feb 01, 2007 6:53 pm
by Mihai
Very strange thing :shock: Could be a scientific discovery of light interaction :) It does look like an interference pattern, light trapped bouncing between the back of the glass and the support.....if you could post a closeup wireframe view...

Posted: Thu Feb 01, 2007 8:11 pm
by tom
I've seen something similar on the edge of a micro-beveled prism as well. But it was rather linear. I wonder if this is a bug due to possible rounding/clipping problems of values but it's very early to talk about it yet. Adrian, due to the originality of your object I would like to have this scene with a simple plane replacement (of course while keeping the same effect) instead of the original stand if you mind. Let's examine what's really going on here. Thank you.

Posted: Sat Feb 03, 2007 5:03 pm
by -Adrian
Ok i got to run some more tests, and sadly the microbevel on the glass plate seems to be the sole cause.

Image
A quick viewport render with a non beveled block in place, no strange artifacts what so ever.
The beveled glass plate still shows heavy patterns under 100% equal circumstances.

Posted: Sat Feb 03, 2007 5:12 pm
by Maximus3D
Hm, in your 3d program can you remove phongshading from the glass surfaces which are close together and try again to see if that makes a difference ? just speculating a little here..

/ Max

Posted: Sat Feb 03, 2007 6:18 pm
by ivox3
...still pretty cool Adrian. ;)

Posted: Sat Feb 03, 2007 6:40 pm
by tom
Adrian, I don't remember if you're working with 3dsmax but in my own
test I identified the problem clearly. It's due to high smoothing angle with
long triangles which could be a plugin problem. Typically 30 degrees
should be fine for smoothing surfaces properly. But when I make a similar
scene (with microbevels) in 3dsmax, the plugin exported the smoothed
microbevel glass with 60 degrees smoothing angle which could be also
due to 3dmax's internal default angle. So, I don't put the blame directly
on the plug-in here. We will investigate it. (see at the bottom of this post)

When I render directly from 3dsmax, the output comes like in your example.
Causes: Proportionally long triangles with a high smoothing angle (60).
Image

When I disable smoothing for the object in max (of course while keeping
the microbevels) it comes as expected. Another way is exporting the scene
as MXS and opening it in Studio and changing the smoothing angle from
60 to 30 and "Recalc" normals.
Image

Of course the user shouldn't worry about above issues and use these
workarounds to overcome this problem. I'm not sure if we can avoid
ringing in these specific geometric conditions with improper smoothing
angles in the core but there could be a solution with the plugins' handling
of smoothing angles when using host applications' primitives.

Thanks for catching this and I will let you know when it's solved. :arrow:

-edit: Yes, now I added a Smooth modifier at the top of the stack which is
even disabled and 3dsmax exported the object smoothing correctly now.
I'm sure our friends will fix it with the upcoming version of the plugin(s).
Since it's clear 3dsmax has this problem, I'm moving this thread under
3dsmax section. Once again, thank you.