Help lighting large area

[alexthg]

i always have problem lighting large areas like this... It's an auditorium hall, with lots of downlights.
i don't think placing hundreds of omnis/ieslight would be the best solution (though it may produce the best result).
I've tried using single plane vraylight placed a bit under the ceiling, but didn't get good results...

anyone care to share some tips?

[studioDIM]

You could turn the light niches into vrayLight material, parting from direct illumination to use GI.
In that case, the number of lightpoints wouldn't change the speed of the rendering.
Of course, forget the same accuracy of directly traced lights, unless you go for something like QMC/LC with good primary QMC sampling.

This is one way. i'm sure others will know other ones

Lele

[Clifton Santiago]

Normally in this situation, I don't put down lights for every light fixture, but maybe every other one. Also, to compensate I make the light cone much wider than it would normally be.

[alexthg]

Thanks Lele & Clifton

with vraylightmtl, i got lots of splotches all over,

so,

I ended up using the 'brute' force... arrayed lots of max spotlight. Pretty much like Clifton's suggestion
And it's not as bad as I previously thought... and it renders quite fast too, as long as i just use standard max spotlight, not IES or Vraylights.

[TutanMark]

Sometimes i prefer to use Vraylights for that situations. Dimentions of those lights could be different to specific areas.

Of course u can use instanced photometric downspot, but... remember tht your rendertime will increase.

Marco

[studioDIM]

with vraylightmtl, i got lots of splotches all over,

You get splotches using QMC/LC as GI method?


Lele

[studioDIM]

I made a quick testbed to verify that, and yeah, you're right about splotchiness.
However, i found out there is actually a better, and much faster, way to do this.
The scene is composed of a plane, a bulgy torus knot, and 16 VRlights looking down straight on the plane.QMC AA 1/4, 1subdiv for QMC GI and VRLights (hence the pronounced noise), RQMC threshold 0.007,Adaptive amount 1.0 for all of the images below.

Direct computation, QMC GI + LC,1m47s on my crappy pc:

LC-stored direct lighting (under the LC options, not the lights' ones), QMC GI+LC,1m31:

Difference in PS Direct-LC_Stored:

Extremised version of the above:


The direct computation image is to be used as a reference for the shadow pattern, as even if noisy, the shadow pattern is correctly calculated.
16 small, far away lights and the bulgy, overlapping object result in broad, faint shadows overlapping. Tricky at least for the LC.
Nevertheless, most of the difference is attributable to below-threshold areallight noise.
There is no VISIBLE difference in the image, and much less so the horrible splotchiness that came out with my first suggestion.
With the increase in lights numbers, direct computation would likely result in insane rendertimes, whereas the LC would suffer minimally, by comparison.
Apologies for that, i hope this makes it up

Lele

[alexthg]

Really thanks for the explanations Lele....

just one question (might be a dumb one)... can I achieve the same result by using Irradiancemap+LC instead of QMC+LC ?