Light Leaks? Where did I go wrong?

[MegaPixel]

Screen Shot:
http://www.cgvisions.com/temp/light_leaks.jpg

Settings:
http://www.cgvisions.com/temp/light_leaks_settings.jpg

Now I've renderd this scene a few times before without these light leak phenomenon and the scene settings havn't changed hardly at all since, so I'm guessing it's got something to do with my recent render setting changes. Could someone take a look-see and steer me on the right path as far as appropriate sampling/IR map settings for this scene? My target resolution for this render is 3072x1536 equirectangular. I was trying to lower my IR map quality down a tad since I was rendering to a higher res. but I think I missed the boat?

Also, regarding Distributed Rendering... Would messing around inside the VFB to make color corrections and such before the rendering is complete cause the render/MAX to crash? This is what happened to me this morning. I have been having the worst luck with Distributed rendering lately. Is DR the only way to use multiple resources to create one image when using Vray GI or is it just IR mapping?

Thanks again - Dave

[Da_elf]

simple enough, your using a preset that has GI blur

[ene.xis]

easy...blur GI!
turn it down to 0

[MegaPixel]

Hmmmm, didn't even see that little area - Thank you both evry much. I'm not even sure why I had the animation preset on - could have swore I hit medium So if I decide to use the animation preset, then I'm agreeing to get light leaks in my scene then (anything with GI blur)?

Any other settings that look rediculous to anyone for what I'm trying to do?

Thanks - Dave

[ene.xis]

hi Dave:
Blur GI purpose is to get smother results with animated objects...when ur not anymating objects u can just turn that value to zero...

i don t think i ll be able to help u much with ur GI solution cause my workflow is now strictly based on OMC+LC...

[MegaPixel]

ene.xis,

The only reason I am using the IR/QMC approach is because it's the first technique I learned. I'm certainly open to rendering this scene out differntly, especially if it means better quality. Are you referring to Vlado's universal settings then for your workflow or?

Thanks - Dave

[ene.xis]

i ve sorted my own aproach..and still haven t tested in prodution vlado s...(am still hoping for 1.5 to do it)...
i just use qmc as the primary bounce and Light Cache as the secondary...
if ur willing to waste some time testing i d advise u to, if ur in a deadline i might wait a bit...it s ur call

[ngrava]

No, you just need to turn on "Check Sample Visibility" in the IRMap settings. Also, you have issues if the corners aren't actually connected. Like, if you have one plane as the celling and another as the wall. Especially if they extend a little bit past each other.

-=GB=-

[MegaPixel]

Please tell me current GI technology is beyond the whole "non-connected geometry" issue? Most all models that I receive these days are made of disconnected - BIM type models where walls/ceilings/floors are seperate objects. Who has the time to remodel everthing as a perfect solid model just so the GI renderer won't freak out?

Just a little frustrated venting there - sorry

[ene.xis]

are there any gaps?
as long as they r intercepting u r just fine...i m almost sure the problem there is the blur GI...

[MegaPixel]

No there arn't any gaps - all geometry meets perfectly flush on all edges and I've rendered this scene out before without leaks so I'm pretty much convinced it's the Blur GI box that did it. I rerendered the scene last night and it turned out fine. Thanks again ene.xis

Dave

[ene.xis]

[ngrava]

Please tell me current GI technology is beyond the whole "non-connected geometry" issue? Most all models that I receive these days are made of disconnected - BIM type models where walls/ceilings/floors are seperate objects. Who has the time to remodel everthing as a perfect solid model just so the GI renderer won't freak out?

Just a little frustrated venting there - sorry

LOL! Yes, current GI technology is beyond this problem. The solution is called "check sample visibility."