Black marks in Mental Ray Interior

[whitewash]

Hi chaps

 

I've been having a crack at an interior scene as a test and having eventually taken the plunge with Mental Ray I've been really impressed with what it can do - I simply can't understand why there's so little documentation for it out there to show folks the ropes. In fact, I think everything I've learned about Mental Ray in the last couple of weeks has come from either here, CG talk or Jeff Patton's website.

 

Anyway, I'm coming to the stage where I'm just trying to get as smooth a render as possible, but keep getting ugly black bits and noisy shadows but not 100% which parameters I should be tweaking to correct this.

 

So, settings:

 

1 Mray area light (just outside the window)

Area light shadow samples upped to 9 u and 9 v.

GI photons 2 million

FG - 300 samples (pre calc, no radius)

Decay is 1.6 (i think)

Energy is default (again, I think)

 

Upping the number of GI photons from 500,000 to 2 million doesn't seem to make much difference to the image apart from dramatically increase render time. Same with FG samples, only more so.

 

I'd like to have a finished render by Wednesday and around other work I've only got a few limited slots for a heavy render like this, so a nudge in the right direction would be greatly appreciated.

 

Cheers

 

Paul.

[vertexART]

Hi there Paul,

first to say, no need to increase the number of GI photons so dramatically while remaining the number of FG samples at 300..?

 

Try this and report

FG 1000;

GI photons 20000;

you can use 'Preview(without calculations)' which will give you a pretty close picture of the overall effect.

 

..then experiment with light decay also

 

For the rest of comments, your light doesn't 'fit' with the rest of background, light shadows are way strong for that time of day, try using:

sky light : multiplier 1.0 with some dark blueish tint

area(direct)light: 0.6, with shadow opacity around 0.4

 

Note that those are wild guesses from experience, I cannot make a test now, I am rendering some work.

Hope it helps and good luck!

[whitewash]

Well, thanks for the tips.

 

I tried the above settings and although they've smoothed out a lot of the noisy shadows (and knocked 30 minutes off the render time) but I've still got my mucky black marks. A few more tests show that the marks are definitely created by Fgather (see attached render with no FG), but playing with the FG settings - increasing and decreasing number of samples, changing sample radius size, increasing filter no. (what does this do?) - doesn't seem to get rid of the black splotches, although changing the sample radius does seem to help a bit.

 

Any ideas?

 

Cheers

[whitewash]

...And since I'm asking, what's the difference between changing

 

Maximum no. of photons per sample (both GI and volumes)

and

Average GI photons per light?

 

(this is in Max 7 - I think the MRay indirect illumination menu is slightly different in Max 6)

 

Cheers

[William Alexander]

Well, thanks for the tips.

 

I tried the above settings and although they've smoothed out a lot of the noisy shadows (and knocked 30 minutes off the render time) but I've still got my mucky black marks. A few more tests show that the marks are definitely created by Fgather (see attached render with no FG), but playing with the FG settings - increasing and decreasing number of samples, changing sample radius size, increasing filter no. (what does this do?) - doesn't seem to get rid of the black splotches, although changing the sample radius does seem to help a bit.

 

Any ideas?

 

Cheers

 

Up the photon count and or radius, untill the GI lighting covers the scene more evenly (less splotches), certianly better than the last image, NO FG in that calc/render. Then process the final fathering with the saved photon GI, start at about 200 and work up in 100 or 250 increments until things even out. I would very much recomend experimenting with the sample sizes on a simple scene with one object that has fine modeled detials, indents, reliefs........ this is where the FG max min settings really show thier value and behavior.

 

 

As far as the scene, a decay of 1.6 really distorts the actual fall off of the illumination, instead 2.o default- adjust the energy levels.

 

WDA

[joske]

the spots near the window wall are from area sample settings

increase until gone 10-10 ... 15-15 ... 20-20...

will increase rendertime of course

 

the spots near the chimney are from bad photon distribution/area settings

you should optimize these untill you have an almost 'finished' render using only GI photons, and no FG

 

area should be between 30-90 cm. (check units AND system units to know if your scale is correct, your furniture should be real life size for instance)

n° off photons should be increased untill result is smooth (usually between 500000 - 2000000, but there is no exact n° to predict, because this will largely be dependent on area size)

 

then just add a small n° of FG 50-2500 if you need more your photons are not setup correctly

the low FG will also give you alot faster renders

 

hope it helps, but the only real advice i can give you is that you will need to practise alot and familiarise yourself with all the settings. and that will not happen overnight ...

 

good luck

[whitewash]

Up the photon count and or radius, untill the GI lighting covers the scene more evenly (less splotches), certianly better than the last image, NO FG in that calc/render. Then process the final fathering with the saved photon GI, start at about 200 and work up in 100 or 250 increments until things even out. I would very much recomend experimenting with the sample sizes on a simple scene with one object that has fine modeled detials, indents, reliefs........ this is where the FG max min settings really show thier value and behavior.

 

 

As far as the scene, a decay of 1.6 really distorts the actual fall off of the illumination, instead 2.o default- adjust the energy levels.

 

WDA

 

Just put that image up to show that with no FG, there are no black splotches on the window frame, which is really what I'm trying to clear up now. I'll have another play with energy vs. decay tonight though.

 

Cheers

 

Paul.

[whitewash]

the spots near the window wall are from area sample settings

increase until gone 10-10 ... 15-15 ... 20-20...

will increase rendertime of course

 

 

Will definitely try increasing the sample settings tonight, cheers. I thought 9 would be the maximum for some reason... oh well.

 

As for the other stuff, I've tried to set the scene up mostly as you say. This thread:

 

http://www.cgarchitect.com/vb/showthread.php?t=8334&highlight=mental

 

probably taught me more about using mental ray than anything else on the net (or anywhere else, for that matter) - particularly the idea of working with and increasing GI photons until you've got a decent picture, then adding a bit of FG after. Unit set up in the scene isn't perfect - I'd already built most of the scene before I came across the above thread - but is now pretty close, and consequently I've been using a GI radius of 60 - 75 cm.

 

BUT - I don't think any of this will help me get rid of the black marks on the window frame, which seem to be put there by final gather. Am I the only who gets these splotches?

 

Cheers

 

Paul.

[joske]

mn... didn't see them until now.

snorry about that

try to increase the max radius in fg 1m - 5 m - 50 m -100 m...

see if that helps

if not try with different material for windowframe

looks more like a raytrace problem, but i'm not that sure. try with a default gray material to see the diff.

[whitewash]

Aha! FG max radius of 10m finally did the trick. Here's the 'final' image - for the time being, anyway. Right now I'm tired and off on my holidays.

 

Thanks for all the advice!

 

Paul.