ShaunDon

Hey guys --

I'm having a rough time with Viz4 ... this is my second time constructing an interior using radiosity -- oddly enough, the problems I had the first time around were minimal. However now the daylight system is bleeding through the ceiling and down the walls, creating a very visible error. Here's my shell:

http://users.adelphia.net/~shaundon/glitch.jpg

I figure this has to be fairly common, but I haven't found an answer anywhere... and the other Viz guys at my work don't have any ideas either. :-( Thanks for your help, guys.

ShaunDon

ZepSOFD

Shaun,
Would it be possible to post a link to the file or at least a striped down version of the file for review?.

tom

ShaunDon

Sure thing. This is the same shell that I posted a rendering of. I hope you'll forgive me, but I wiped out the radiosity solution 'cause I don't have that much web space -- but I haven't changed anything in the file, so you can recalculate it and it should be identical.

http://users.adelphia.net/~shaundon/...erson_east.zip

Thanks for your help.

ShaunDon

ZepSOFD

IQ: 80 -- Mesh El.: 37509
RI: 30 -- Rad. Sol.: 3min.27
Fil.: 1 -- Ren.: 5sec.

Shaun,

Here are some things I did to remove some of the light leaks:

1. Rendered with direct light to avoid large mesh in getting shadow detail.
2. Model ceiling for each room. Using one ceiling across intire project with out breaking the face will almost always generate light/shadows leaks.
3. Model floor for each room. Same as 2 but sometimes you can get away with this in VIZ if you can afford a fine mesh for the floor(this is strictly scene dependant) but you would probaly be better off just breaking it up. I didn't break the floor up in the above image and I did get light leaks below the walls. Also most of the time it is best to break objects up into multiple material areas instead of overlapping faces like in the red area of the floor this will most likely cause light/shadow leaks or some kind of rendered artifact.
4. I lowered the walls to the bottom of the ceiling where I marked in blue which eliminated the light leaks on that wall but if you look to the left I did not change and they leak. Before I lowered the walls I was able to reduce the light leaks to a level I thought was ok, but the areas above the doors just based on how they were modeled I could not reduce at all so my next plan was to either break the wall at the ceiling line or lower the wall, I took the easy way out and lowered the wall. I also did not cap the wall and on the right side side you can see where the light went right through the face and ended up rendering in the scene.

I hope I am not being a nuisance or overstating the obvious.

Good luck,

Tom

ShaunDon

Tom --

Thanks for takin' the time to sort through that, man. It never occurred to me that it was the overlapping that was the problem... that's a safety margin that everyone I know always tends to leave. Course it makes perfect sense now that I think about it... the walls were extending above the ceiling, so of course they'd recieve that light... and most 3d programs have issues determining intersections. After years of AutoCAD that alone should have occured to me.

The way the doors were modeled was my own work-around to try and solve the only problem I encountered in my first rendering -- I had three large windows, but for some reason the light was only penetrating one of them. I figured the only thing that could possibly be causing a mesh error were the boolean operations -- but as soon as I rendered my first radiosity solution with this model, I saw that by avoiding booleans I was dooming myself to shading hell. Something I just have to deal with.

Thanks for showing me how you fixed this stuff -- it'll be great to know for the next time. As of now, I'm lighting it all with standard omnis and radiosity calculating the interior... so the light's distributing fairly well, I just miss the mix of the daylight. Oh well, the customer won't really notice. Thanks again very much for your help

Regards,
Shaun

ZepSOFD

Hey Shaun,
I'm just glad I can help in some little way. This is mainly the problem with radiosity, if you were just using scanline you would still be happily ren, ren, rendering along.

Take care,

tom

STRAT

i've encountered this light leakage problem with walls/ceilings running thro in 3dstudio's scanliner for years now.

__________________

ZepSOFD

It's been just the opposite for me with scanline, occasionally I'll get some weird artifacting but generally I find it to be very lenient with acad models. Oh well go figure,

[ September 18, 2002, 02:33 PM: Message edited by: ZepSOFD ]