CompuServe Thread

#3DS LIGHTING

4 messages in this thread
#33672From: Paul BrinckmannJan 28, 1993 10:33 AM
Here is a very technical question for you guys. We have a customer that wants as close to reality of indoor lighting as possible. They need to know what brightnesses and positions to put their lighting at in there building. Are we able to set the manufactures foot candles and light temperature color to each individual light source, and if so how? I believe we are looking for a math formula that would convert the variables to 3DS values… and how accurately will 3DS render the image.
#33677From: Yost GroupJan 28, 1993 10:43 AM
If you're looking for a physically-accurate light model, the ONLY rendering method that will suffice would be a radiosity system. There are a number of these running on the SGI, but I don't have any names handy for you. – G
#33698From: Paul BrinckmannJan 28, 1993 3:19 PM
Gary, What we were hoping for was a best guess way to do this in 3DS since we are hoping that in the near future versions of 3DS will far surpass that of SGI and we wouldn't like to off toward the idea of using SGI.
#33926From: John DowdellJan 31, 1993 8:05 AM
One problem with modeling accurate indoor lighting using the current generation of 3D packages is that of diffuse inter-reflection of surfaces. Ray tracers handle specular reflections well — they can figure out how light bounces from shiny surface to shiny surface — but they break down when trying to figure how non-shiny surfaces reflect light. If you look up at the ceiling, you'll probably see that a corner of the ceiling is subtly darker than where the wall and ceiling meet in the middle of the room. Fewer diffuse inter-reflections there, less light. The same phenomenon won't show up with our current computer light models. Like Gary said, radiosity reproduces this phenomenon. The only PC package I know supporting it is StrataVision on the Mac. Calculation times are horrendous though. Some of the Amiga guys simulate radiosity by using multiple soft point light sources, but there goes accuracy out the window. Would your client accept an *estimate* of light needs, instead of decimal-point accuracy?