In working to make a realer looking florescent light for my hallway project I've changed the Light model in the hallway. I created 3 upward facing keligs to illuminate the inner metal of the Light frame. However, when ever i go to render radiosity it creates a leak in the object and radiosity rays appear to go strait through the metal.

It definatly does not look real, and would realy bring down the believability of the renders. Any ideas of what this could be? or better yet how to fix it? lol

Is your metal only one patch thick?

Nope. It's very much a solid. Would that effect it if it was? Any other possibiliteis?

I am a nut without radiosity experience... but...

Is it possible the lights are "too close" to the metal frame? If they intersect the mesh maybe just maybe... the light might leak through.

Check normals... probably not the problem... worth a look. I always get some normals "backwards" with shapes like that.

One more idea... the mesh of the light frame looks.... "lite" to me... very sparse. maybe it needs a few more patches.

I only mention these things because they can sometimes cause problems WITHOUT using radiosity.

Vernon "!" Zehr

Null.

Well, i checked the model for reversed normals. They are all set correctly now. I did check to make sure the lights were not cliping through the metal housing, and they were not. The Shadow example was to test for any light leaks of that nature... i turned them up to 1000% to check for any leaks at all and there were none, but with the normals and lights checked the problem is persisting.

Anyone else have any idea on what this could be? I'm stumpped. I've tried loweing the lights, lowing the housing, changing falloffs, lessening klieg angles, raisng klieg angles, changing the boxes reflectivity properties, asking the dog... it's just not solving this leakage problem. I sure hope this can be fixed.

What are the settings on your kliegs? What kind of shadows do you have them set for? Are the shadows set to 100% dark? I am pretty sure that one of the shadow types available to kliegs is a no-no for radiosity.

And to answer your question about the effect of non-solid geometry - It makes a definate difference with lights. In my personal experience, kliegs will bleed through single patch geometry in a very strange way. You will get glows on the back side of a single thickness wall with a klieg on the other side. Shadows can appear where there is no light (somehow the light got blocked but the shadow continued)

Bill

To give you informed advices, you also need to show us how your lights are positionned and setup inside the reflector.

Photon Mapping is a statistical radiosity technique which is why it is so much faster than pure Monte-Carlo techniques. To compute the illumination at a given point, it averages a bunch of photons in the neighborhood of that point. This neighborhood is determined by the photon sampling area. Thus the thinness of the geometry that it can handle correctly is proportional to the photon sampling area. 1/4 the sampling area to be precise. 1800 as sample area (18cm radius) is very large. That means that it will potentially catch photons that are up to 4.5cm away on the other side of the light fixture (that would be inside the light fixture in that case). In normal situation that is not a big problem and enlarging the sample area will just blur the effect of those misbehaving photons. But in your particular case, you have a very high density of photons with high energy packed inside the light fixture so they contribute a lot.

There are two solutions to this situation:

1) Increase the number of photons (probably to the max number of 1M) so that you can reduce the sampling area.

2) Don't shoot your light directly inside the fixture. I recomend this solution for two reasons : a) By shooting the light directly inside the light fixture, you pack a whole bunch of your available photons in the fixture for nothing really worth it leaving less photons to actually illuminate the scene. Even if you solve the light leakeage behind the fixture, the final gathering will have trouble and generate noisy images with very bright white speckles because the fixtures will be much brighter contributers than the rest of the scene. You will then need to increase the number of final gathering samples and will still have noisy renders.

The correct way to setup your lights on this fixture is to have an array of klieg lights with 170° angle, a very small width softness, turned straight down and positionned inside the fixture is such a way that it does not directly illuminates the inside of the fixture. 3 or 4 klieg lights per fixture should do the work.

Here is the Rig for the florecent box

Klieg Up:

width: 30cm
width soft: 100%
fall-off: 50cm
cone angle: 160
RayCast: 2
Shadowdarkness: 100%

Ok, I made the box much thicker, approximately 1.6cm thicker (any thicker and it's gunna look like a cartoon), and i raized the photons caste to one million. It looks pretty good now... but there is still some light comming through.

Radiosity Settings
Photons Cast: 1000000
Sample Area: 1800
Photon Samples: 500
Intensity: 100%
Max bounces: 15
Caustics: OFF
Final Gathering: ON
Samples: 200
Jittering: 0%
Precomputer Irridecence: ON

I renderd a strip above the fixture to show what it's doing still. It's lookin much better though. I may just keep it like this. Thanks guys for all your help. If you have any further suggestions plz let me know.