Why the caustics so low in small mesh light

[KSP_STAR]

As you can see, when the mesh light gets smaller, the windows caustics get smaller.

Higher samples can't handle this problem (I tried more than 50 samples to present it, but it didn't work)

The light intensity is the same

C4DtoA 3.1.0

[lee_griggs]

What exactly is the problem? Because you reduced the size of the light? Can you upload a scene somewhere?

[KSP_STAR]

yes.The windows caustics problem.

[KSP_STAR]

caustics.zip

[lee_griggs]

Try increasing the number of mesh light samples.

[KSP_STAR]

I have been try this way,but no use...

i used 50 samples to render it...

[maxtarpini · Accepted solution]

Oook 🙂 I had a minute to test it.

The problem is what I already described in the other comment.

But that only applies if you wanna deal with real caustics. You should not have that kind of problem with default mesh lights as it looks arnold resorts to direct lighting with simple transparency to deal with this kind of scenario.

So when you deal with real caustics (specular-diffuse paths) in Arnold ?

When you use emissive materials. Or when you use visible mesh lights.
(I'm lying here because there's also of course specular contribution from arealights 😉

Turn off visibility for your mesh lights and you'll deal just with transparency. Maybe replace the visible lights with some cube/plane/etc with very low emissivity that won't reach other surfaces.

Eventually, if you are curious you can disable diffuse light contribution to just see specular contribution (aka caustics). Here in maya. (click the image to see it full res).

Here 5000 diffuse samples with emissive material. Not quite there.

Simple transparency.

[KSP_STAR]

Thank you very much

[Pablodeleo]

The light with geometry is more heavy. You can liar the light with self shader and lighting hidden is more easy and optimization for now. In the future maybe have a solution bether.