Strange shadow computing in corners

[PIXELBOX_SRO]

Clifton:
tried both methods u have described...doesnt work

tried presence of any unwanted material
tried applying med grey to everything
tried checking sample visibility

NOTHING HELPS...((((((

[PIXELBOX_SRO]

uploading to rapidshare now for anyone willing to give it a try....will post the link soon

camera am testing is no.6

all lights apart form sun hidden
setting might be very skirmish as ive tried all sorts

[PIXELBOX_SRO]

i think that if it was a geometry related issue, i would not happen on the newly created solid boxes in front of the camera...dont u think?

i also have a dome light mapped with hdri in the scene but that light is off and hidden

[PIXELBOX_SRO]

finally...here is the download link

http://rapidshare.de/files/40479000/test.max.html

big thanks

[PIXELBOX_SRO]

so far ive found out the problem starts with LC...the smaller sample size , the bigger splotches.....increassing IRR settings smoothes this out though...but still its the first time this has hapenned to me

[Clifton Santiago]

What Max and vray version is it? I might be able to test it out tomorrow.

[PIXELBOX_SRO]

its max 9 64bit , vray version...the latest official one 64bit

really weird this...ive also tried setting diff GI ID for those intersecting objects
no joy...the only way is to set the LC sample size to 0.01 and subdivs to 1000 and really high IRRAD settings like 70/40...madness

[PIXELBOX_SRO]

but then again...should it not be getting better with more detailed LC setting than the other way round?...ive read some posts here such as this one:

http://www.chaosgroup.com/forums/vbu...ht=light+leaks

and this one

http://www.chaosgroup.com/forums/vbu...t+leaks&page=3

but for me only above written works better, still not very happy with the GI detail though

[RErender]

I don't think it is the LC per se, I tried it with the secondary bounces turned off and the splotches were still there. Changing the sample lookup to Overlapping helped a bit but of course that is just blurring the solution a bit more. I've run out of time to look at it with no real solution...

[PIXELBOX_SRO]

thanks rerender for your effort...at least i know am not completely dumb by leaving some important checkbox unticked

[Morbid Angel]

I don't think it is the LC per se, I tried it with the secondary bounces turned off and the splotches were still there. Changing the sample lookup to Overlapping helped a bit but of course that is just blurring the solution a bit more. I've run out of time to look at it with no real solution...

I didn't know we had time constraints

[RErender]

heh, got other stuff to do

[jacksc02]

just downloading your file now to have a look.

[PIXELBOX_SRO]

thank you guys, appreciate your help

[Morbid Angel]

here is result:



settings were:

irmap:

-2/0
80/20

LC:

1000
screen size 0.02
prefilter 30
interp samples 20

thats it.

Typical resulotion for such scenes: take some steps back. First thing I did, was to switch both gi's to light cache ppt mode. After few minutes I saw that there were a clean gi solution. That lead me to believe that there was a problem with an irradience map/lcmap setup. I then tweaked the existing settings somewhat. Mostly the section of -2/0. That is important when you have such large indirect bounce; a precision of gi is affected. By setting irmap samples to be -2/0 you ensure the precision and thus remove error.