I normally think of radiosity as a way to mimic real world lighting, but after rendering out a recent series of images using light rig radiosity, I noticed that the bottoms of objects were much much brighter than the tops. Is this an accurate depiction of real life, or is this an inherent problem with photon mapping?

It looks like an exaggeration of real life.

Doing still lifes in art classes we would lighten the underside of objects to depict the light bounced off the table top.

But not that much!

maybe there's a setting that's off?

After doing some other tests, i'm sure that it's a limitation of photon mapping, and not any problem with my settings. Photon mapping is the creation of an extreme amount of random light rays, and i think it's more likely that a ray is going to get stuck under the sphere than bouncing off the dome walls and onto the top and sides of the sphere :|

If that's a typical result of Photon mapping then Photon Mapping isn't much use. There must be an arrangement that gets better results.

there's a lot of things that could be going wrong. There's a ton of lights, and that could be casting an enourmouse amount of light towards the bottom of the sphere before it even hits the ground. The statistical analysis done by photon mapping is most likely the cause. Still, it is much faster than monte carlo radiosity; looks better too.

Indeed, there is a ton of things that can explain this. But in the absense of any information on the said "light rig radiosity", it is totally impossible to make informed comments and suggestions.

Yves: I used one of your skylight rigs inside of a white hemisphere and then renered it with radiosity. The idea is that photon mapping will mimic ambience occlusion -- with rays bouncing off the dome walls from all directions onto objects in the center -- but also have the effects of radiosity.

I now think that it is the object's size which is creating the harsh contrast in brightness. I increased the scale of the sphere to 400%, which gave a much better looking result. From these results, I believe it is safe to say that if an object is fairly small, fewer rays are going to strike the object's surfaces which look towards large open spaces, and more rays are going to strike areas of the object which are close to and look towards larger surfaces, like the ground, where there is much more photon activity. I guess a good analogy here would be someone trying to measure how much rain fell in a day but they only used a thimble to collect the rain sample. He or she is not going to collect hardly any drops at all with something as small as a thimble -- though, if on the ground, its base is bound to get wet. So, with a smaller sample, the measurment will be very innacurate, and it would probably be better if he or she used a much larger collection device, like a bowl the size of a radar dish.

Hm, I hope that makes sense. I am not too good with analogies, and I could be dead wrong here again

Ît seems odd to me that you would use a light rig along with radiosity. As you write, the skylight rig will give you an AO effect. A sort of fake radiosity. And then you add radiosity to that? That looks like a cool experiment but hardly usefull in an practical application. I would venture that you would get just as weird lighting results if you were to use such a light rig on a white ball in a real closed white room setup.

Well, i guess it's all in how someone uses it, or what the personal preference is. AO is really nice, but aside from generating a full 360x360 light, it has no radiosity :|

Here's two renders. The one with the black background was done with ao, and the other: radiosity.

hey umm just want to make sure this was right looking as well
ive always used HDRs for my renders but not with an actual HDR background
just was wondering if this looks right

It does look like it's in that environment, so I would say it looks correct.

By the way, if you want the environment to be reflected in the metal bits, you'll need to add an environment map.

evierment map to what?

To the metal bits.

well, if he's already using a dome for the background, normal reflections should work fine. Trying to get both the env map and the dome to match up is brutal.

yeah that has been a problem but if i just rotate around after maping it matches just fine
still can quite figure out the enviorment map thingy for the reflection