PDA

View Full Version : Is there a limit to how many photons you should use in a scene? Mental Ray


MisterS
11-30-2010, 01:47 PM
I have a model of an underground chamber lit by a daylight system focused on the opening of the chamber.

GI is emitting 100 million photons, producing a 2GB map. I need to emitt a significant amount more though but can see my system resource chugging away. I saw memory go up to 7GB out of 8GB so am thinking i'm at the limit.

I'm keeping animation in mind so would like to keep the photon map rather than use Importons to ditch most of it (Importons are quite new to me btw).

Is there a better approach?

Brian-Bradley
12-01-2010, 08:02 PM
Hi David,

That is quite a big memory hit for the photon map! Without seeing your scene the
first question I would ask is 'How come the Photon count needs to be quite so High?
What sort of set up are you using? (photon settings that is) In fact what host app
(3d package) are you using?
The more info you provide, the more help forum users will be able to provide. :)

Regards
Bri

MisterS
12-01-2010, 09:06 PM
Hi

I'm using 3DS Max with mr sun/sky.

I tried again emitting 500 million photons using importons and it worked, map was around 70MB which is quite cool. The results were ok too, a fair bit of illumination, I'd still be tempted to try a higher setting. I left it on overnight so not sure how long it took but I think it prob would have taken a good couple of hours. PC didn't wake up in the morning - had to restart but it loaded back on my orig scene - Result!

Think needing this amount may indicate I need to try another workflow.

I'll post some screen shots when I'm back in work on Friday.

Since my original post I've been watching a few tutorials and as far as I can see the problem is the fact that there just isn't a great deal of light entering the chamber hence the need for loads of photons. I should have also used a mr sky portal on the entrance hatches to my chamber also which may help.

I'm using a mr arch design material, mid grey and matte. Wondering if I need to cheat and put some lights inside the chamber.

The model is of an underground sewage pump station that my company is refurbishing.

Brian-Bradley
12-01-2010, 09:31 PM
I'm using 3DS Max with mr sun/sky.

Hey David,
Okay, Max user myself so the first question I would ask is are you setting
the Photon 'target' Radus to your opening size? (on your MR Sun that is)


I'll post some screen shots when I'm back in work on Friday.

Cool.


Since my original post I've been watching a few tutorials and as far as I can see the problem is the fact that there just isn't a great deal of light entering the chamber hence the need for loads of photons. I should have also used a mr sky portal on the entrance hatches to my chamber also which may help.

I'm using a mr arch design material, mid grey and matte. Wondering if I need to cheat and put some lights inside the chamber.

The model is of an underground sewage pump station that my company is refurbishing.
Okay, got a better idea of what is going on in there now! :)

definately use a sky portal where you have very limited openings! (turn of its shadows if
you do not need them and your portal render times will go Waaaay down).

Light cheats can be used, but you could also up the reflectance value of your material
to cheat a little extra light in there!

How many bounces do you have on your photons? Obviously this will determine how
much light they are carrying around to space!

Regards
Bri

MisterS
12-01-2010, 10:04 PM
Yes I have set the photon target radius. I will try a sky portal, not used for a while - but think I will def need shadows. There are lots are shadow casting nooks and crannies.


The materials will all eventually have some reflectancy.

RE: bounces for photons - not sure what you mean. trace depth is at default of 10.0. Decay is also default of 2.0

I also used FG with no bounces. I thought this was best when used with GI.

RE faking lighting - I think this is somthing I should maybe learn.

In particular I'm trying to get some detail inside the chamber. I have switch AO on in the A&D shader but again, this is on default. Some of the objects still appear to be floating.

A tutorial I saw this week suggested an AO pass - need to learn how to composite then too!

Brian-Bradley
12-02-2010, 08:39 AM
RE: bounces for photons - not sure what you mean. trace depth is at default of 10.0. Decay is also default of 2.0
Hey David,
Trace Depth essentially controls how many times the photons will 'bounce' in the
environment. Try setting it higher but with a lower number of Photons and see if it
gives you more light but with less memory hit!

I also used FG with no bounces. I thought this was best when used with GI.

As soon as GI (photons) are enabled FG bounces become automatically disabled anyway
so a 0 bounce setting in FG is all you will get. :)

Regards
Bri

MisterS
12-03-2010, 10:59 AM
Attatched is a render using 500 million photons and FG. Model is still WIP.

I plan to block up the large pipe that you can see coming into the chamber as there would not be any light comming in. I'm hoping to eliminate the floating objects, particularly the baffle wall connected to the edge of the chamber.

Will try your suggestions later today.

InfernalDarkness
12-19-2010, 08:52 PM
Cool scene! Looks like a fun lighting experiment as well.

Honestly, given the nature of this scene, I don't think you should be using GI at all. Seriously - there's no need for photons unless you're going to be shooting caustics everywhere, and this is still best controlled with focused spotlights for your caustic photons.

Give it a try with sun/sky, a portal light, and FG only? FG has come a long long way in efficiency and speed, scaling well with more modern computers as well. I so rarely use photons anymore in favor of portals and FG only - and generally the results are just brilliant.

CGTalk Moderation
12-19-2010, 08:52 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.