View Single Post
Old 18-10-2003, 06:38 AM   #74
Zytrex
I am mad, you know.
 
Zytrex's Avatar
 
Join Date: Oct 2002
Posts: 207
Thanks: 0
Thanked 0 Times in 0 Posts
Awards Showcase
Threedy Top Poster - Silver: Contributing to the forums with 1,000+ posts - Issue reason: Target reached! 
Total Awards: 1
I agree RickStefani. When I first got access to radiosity, I became overwhelmed by how amazingly easy it was to create amazingly realistic lighting and bla bla bla bla. Sure, it's fun to take a box, reverse the normals, put a few smaller boxes inside, add an omni, and then watch the light bounce around, but it doesn't go much farther than that. In any real scene--and I mean <i>scene</i> as in there are lots of objects and light sources and materials and all kinds of things--radiosity quickly becomes a burden. Personal computers are just not fast enough to make radiosity worth it. I mean, am I the only person who noticed that, every time our computers get fast enough to do pretty much everything that our 3d programs have to offer pretty quickly, the software developers come out with a new feature that we end up using and that kills our render times?

My scene is still pretty simple. It started out with no GI and just a few light sources and no objects other than the room. Now it has two glass tables, a chair, and all kinds of complicated multilayered textures. A long time ago, someone suggested I add GI to it. I did. It was amazing. And so I was stuck with it for about a year. Finally, I got sick of the obscene render times. Having to wait eight hours for four frames was just too much.

So, I got rid of the radiosity, messed with the ambient lighting and the decay of the main lighting, and ended up with results without radiosity which were very close to the those with radiosity. Ambient lighting does not light the darker areas a nicely as radiosity, but it still works. Also, radiosity is still a new technology and it can be difficult to get acceptable results. Sometimes objects will be improperly lit on a few faces, or shadows will be a little off where faces of different objects touch, and things like that. These problems still happen with the quality settings at very high and very slow levels. Maybe max 6 will be different, but I'm not about to shell out that kind of money just for a new version of radiosity when I'm very happy with my standard lighting.

The scene without radiosity is a bit darker over all, but that's how I wanted this version. It could have been just as bright, so please don't think that is one of the differences directly caused by not using radiosity. The darkness of the shadows, on the other hand, is a direct result. Although as I'm typing, I realize I could just adjust the shadows so that they are not as dark and basically fake those areas being lightened by bounced light without having to increase the ambient lighting which would throw of the brightness in other areas. Cool! I gotta do that.

In truth, I'm happier with the non-radiosity version. It renders fast and I can use the raytrace material without it throwing the radiosity and render times completely out of whack.

By the way, some of the objects have been moved around a bit and the materials, especially the reflective properties, have been altered. The two images were produced about a month or two apart. Please take that into account when comparing so you don't confuse those kinds of changes as being linked with the lighting changes.

So what do you all think, hmm? It is pretty easy to tell which image uses radiosity and which does not, but only because they are both right there for comparison. Now let me ask you this: If I had only posted the non-radiosity version, but did not label it as such, would you be able to tell for sure if it used radiosity or not? Hmm?

__________________
If you had a neck and I had a hand, I'd squeeze your brain, which is your body, right out of the top of your head, which does not exist!

Discuss Lighting Here
Discuss Subdivision Surfaces Here
Discuss Materials Here
Zytrex is offline   Reply With Quote