More info regarding Omega SSS bug.

[serge]

With Omega-SSS sometimes polygons of the object show up in renders. I've seen many people here having this problem. I guess it's save to call it a bug, since nobody told them what they did wrong.

I don't have the solution, but I did some tests which clearly show when the polys do and don't show up. Hopefully it helps finding and solving the problem.

.....

About the scene/object:

-The Omega object has doublesided polys.
-The 'room' is a simple box with flipped polys.
-Three pointlights: one at the camera, one at the side, and one behind the object.
-Omega full raytracing.
-No radiosity.


The pictures:

1: The object in an empty environment. First of all, the rendertime is very long. Second, the polys are showing (you can see that clearly at the bottom part of the object).

2: Same setup, except I used a logic-node which sets the backside of the polys to 100% transparent. First, the rendertime is much faster. Second, the polys don't show up. Is this the solution then? No, see picture 3.

3: Here I placed the object inside a room. Hey, the polygons are showing again! Even though the backside of the polys are 100% transparent.

4: Same scene, but I set the object-render-properties of the room(!) to "unseen by rays". The polygons don't show up! Weird eh? I didn't change anything on the object itself. This of course is not a solution, since most of the time you do want your environment to be seen by rays.

[serge]

Here's a picture of the node-network, the scene, and the scene in a zipfile.

[Earl]

Hi Serge,

Thanks for posting these. This is quite interesting, and as far as I'm concerned this is a bug with the Omega shader. I hope NewTek corrects this bug in the near future, because until it's fixed I'm pretty much giving up on using it (and otherwise it's a very decent shader, if not a little slow).

However, one thing I noted about your node screenshot - it didn't look like you had anything to set the refraction of the backside polys to 1.0 (it should be set to something else). However that won't solve the polys showing up.

[serge]

... it didn't look like you had anything to set the refraction of the backside polys to 1.0 (it should be set to something else)....
Yes, as you said, it doesn't solve the problem. I just tried to keep the scene as simple as possible.

[jameswillmott]

Copy your Omega geometry into a new layer/object. Set one copy in layout to be unseen by rays/seen by camera, and the other one to seen by rays/unseen by camera. Try that see what happens.

[serge]

...... Try that see what happens.
Nice one, James! The polys don't show up....

However, with this setup I placed a mirror in the background. See the picture: the object renders fine, but it's reflection shows the polys! It's getting a bit complicated now... 8~

[zapper1998]

I have the same problem???

So doing what you did will solve that???

[zapper1998]

Same same...
All I did was change to "Perspective Camera"
The Previuos image was done with "Classic Camera"

[Karmacop]

Increase the penetration distance or subdivide the surface more. I don't think this is a bug, it's just the way the surface samples.

[zapper1998]

Increase the penetration distance??
Ok I am lost..Help.. where is the penetration distance???

Will load a image of the Nodal later...
Thanks
Found the penetration distance....

[jameswillmott]

I've been writing my own SSS shader and am having the same problems, even with a very basic algorithm.

It would seem that a raytrace command simply cannot see the polygon it 'originated' from. I agree that it doesn't seem to be a bug, it's too specific. I guess this is to avoid infinite recursion or something. Just make sure your SSS range is greater than your polygon size.

[serge]

Same same...
All I did was change to "Perspective Camera"
The Previuos image was done with "Classic Camera"
All my examples were done with the Perspective camera, so it's not a solution.

[serge]

Increase the penetration distance or subdivide the surface more.
I don't think those workarounds are reasonable.
I don't think this is a bug, it's just the way the surface samples.
If it's not considered a bug, then sure someone of Newtek (or Tesselator) would have popped in to say what we're doing wrong, no? If they don't plan on fixing it I'd be very disappointed.

[Kurtis]

Have you reported this with your example content to lwbugs@newtek.com?

If you are in the LightWave v9.2 Open Beta, and it still happens there, you can also report it through the Open Beta bug tracker.

[serge]

Have you reported this with your example content to lwbugs@newtek.com?
I haven't. I thought I'd better wait until 9.2 Beta. (Whether it's still necessary to report I'm not allowed to say here...)