Author Topic: IES lights making things extremely noisy / slow?  (Read 13163 times)

2013-10-25, 12:44:08

agentdark45

  • Active Users
  • **
  • Posts: 579
    • View Profile
Hi guys I've been having some trouble with IES / down lighting recently. When adding in IES lights things seem to take 3-8x longer to render due to all of the additional noise.

The 1st attachment shows the scene lit with just an HDRI and an emitter material for the ceiling light, render time ~ 15 mins.
The 2nd attachment has all of the IES lights enabled (photometric light type in max with Corona shadows ticked), render time ~ 40 mins
The 3rd attachment uses a very dark HDRI and all of the IES lights enabled. This one took around 50 mins, you can see the extent of the noise...

All renders are using the same exposure with basic render settings (HD cache, adaptive: 4, latest daily build) One thing that did strike me as odd is that I had to crank the lm values of the lights to 35000.0 else they were too dark.  Exposure is set to 2.75, highlight compression is set to 4.

All glass is set to hybrid mode, the pool is a single plane set to onesided (solid), but with "casts shadows" unticked. Trees are opacity mapped and the RGB of all white materials are set to 180.

Also I'm rendering this on an overclocked 3930k with 32gb ram. I completed the Corona benchmark scene in about 3 minutes 20.

I don't particularly need to use IES lights, I just need some down/spot lights - so anything that can shave some render time off will be a massive help.

Thanks for all suggestions / tips!
Vray who?

2013-10-25, 13:35:29
Reply #1

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12708
  • Marcin
    • View Profile
My suggestion: more path tracing samples (32? 64?) and/or higher lights samples multiplier.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2013-10-25, 13:37:06
Reply #2

agentdark45

  • Active Users
  • **
  • Posts: 579
    • View Profile
Already tried upping both the path tracing and light samples multipliers but they don't seem to help.
Vray who?

2013-10-25, 13:47:53
Reply #3

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
some IES files + area lights can be quite inefficient now, you can for now use 3dsmax photometric lights with IES profiles instead (in daily builds)
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2013-10-25, 13:59:47
Reply #4

agentdark45

  • Active Users
  • **
  • Posts: 579
    • View Profile
some IES files + area lights can be quite inefficient now, you can for now use 3dsmax photometric lights with IES profiles instead (in daily builds)

Ah I see, so rather than using a custom ies file it's better to use a 3dsmax preset one (for example the recessed 250w wallwash)? Should I enable corona area shadows, or leave this unticked?

Would using a regular 3ds max spotlight with corona shadows be quicker? Or would the quickest solution be to create a Corona sphere light inside of a hidden piece of occluding geometry (for example a hemisphere with a shell modifier on it)?

Thanks
Vray who?

2013-10-25, 14:59:02
Reply #5

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
no, you can use any IES. The trick is that 3dsmax lights are always pointlights (even with area shadows), which are easier to compute for nontrivial directional distributions. This should be much faster than either using complicated IES + coronalights, or occluding coronalights with geometry.
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2013-10-25, 15:13:33
Reply #6

agentdark45

  • Active Users
  • **
  • Posts: 579
    • View Profile
no, you can use any IES. The trick is that 3dsmax lights are always pointlights (even with area shadows), which are easier to compute for nontrivial directional distributions. This should be much faster than either using complicated IES + coronalights, or occluding coronalights with geometry.

Thanks for the reply!

So, after some more testing it seems that the ceiling coffer lights are what actually appear to be causing most of the noise! I guess it is a worst case scenario for GI as the lights are hidden in the ceiling and light the rest of the scene via pure bounced light. I might have to fake this one with some invisible rectangular lights around the recessed ceiling borders...
Vray who?