PM Caustics - still having issues

[simmsimaging]

I asked about this once before I think but I'm not sure there was a clear answer at the time. I have always had problems getting decent caustics with water droplets. I'm testing again and so far the best results have come from disabling GI refractive caustics because that was causing some droplets to light up and others to go/stay darker so they looked really uneven. Using no caustics (and affect shadows) works okay, but has the obvious limitation.

Using PM caustics I seem to be unable to get a nice caustic effect *inside* my refractive objects. It tends to be a large soft effect all around them, with the result that the drops always look darker against the surrounding area. I don't think that is correct. If you look at the caustic render element you can see the refracive objects are knocked out of the caustics entirely.

So if want to reduce the soft glowy area around them I have to reduce the search distance, but it just gets splotchy and does not improve the caustics inside the drops much, or at all.

The example below is from a plane that is about 20X30cm, so fairly "real world size" drops. Caustics settings are: search distance .2cm, max photons 0, and max density is .02cm (I have dailed that down from .05 and played with various numbers without seeing a really obvious difference). Only one spotlight is creating caustics and it is set to 10000 sub-divisions.

What needs to be done to get this looking better? Do PM caustics just not show up *inside* refractive objects or am I missing something?

/b



caustic pass:

[Morbid Angel]

whats PM caustics?

[anchovy]

whats PM caustics?

Id assume Photon mapped.

-Michael

[Morbid Angel]

For the raindrops or water drops brute force caustics always give me a good result.

[simmsimaging]

Photon mapped.

The BF caustics were causing other problems so I am back to trying this approach.

[Morbid Angel]

how are you emitting those caustics? from point light or area? Also did you see the thread by vlado about light cache caustics?

[simmsimaging]

In that test it was from a point light (max standard spotlight) but I have tried with Vraylights as well. Currently testing again with Vraylights and GI caustics only. Anyone know if the enable caustics option for lights in the Vray Properties affects GI caustics as well as PM caustics?

I am not sure what thread you are talking about - although I have looked over quite a few threads on caustics I don't recall one about light cache, nor did I find one with a search?

b

[Morbid Angel]

talking about this one http://www.chaosgroup.com/forums/vbu...light=caustics

[simmsimaging]

Ah - the light cache thing threw me a bit I have tried the directly visible approach as well, but on Vlado's recommendation I dropped that approach in favour of using regular caustics.

Here is what it looks like with GI caustics using BF/LC instead of IM/LC and PM caustics: in many respects it's better, but hte light inside each drop is varied in a way that doesn't look right me, and now it is more like all the light is trapped inside - sort of the opposite of the first render test. Both can be fixed a bit in post, but I feel like the PM caustics gives me more even results *within* the drops.

[simmsimaging]

for completeness: here is the rendering with the "directly visible" caustic option on. This is PM caustics, using the Max spotlight with 12K subdivisions. Search distance I lowered to .01cm and Max density and max photons were both at 0

[kosso_olli]

Tested this some time ago, too. I made these using brute force caustics, this looked best.
Also tried the photon mapped version, but I got the same result as Brett: Dark spots inside the droplets and very blurry "rings" of light around the caustics.

To be honest, caustics is the weakest point of vray. Maybe I am tired of fiddling around with the parameters, as sometimes I get completely unexpected results like the dark spots inside the droplets. That is why I am really looking forward to the vertex merging approach Vlado is talking about here: http://www.chaosgroup.com/forums/vbu...Caustics/page5

Looks quite straightforward to me.

Regards

Oliver

[simmsimaging]

I'm sure the BDPT options will resolve a lot of this but it seems like we should be able to do this now. Vray is certainly capable I think.

Can chaos answer a few questions?
1- does the "affect shadows" the material option affect GI caustics? It dies not appear to but what should that be set to when using BF for caustics?


2- does the Vray properties "affect caustics" option for lights affect GI caustics? Again it does not appear to.


3 - is there any reason to expect GI (or PM for that matter) caustics to be messed up by interpenetrating geometry?

[vlado]

1- does the "affect shadows" the material option affect GI caustics?

Yes.

what should that be set to when using BF for caustics?

It should be set to off.

2- does the Vray properties "affect caustics" option for lights affect GI caustics?

No.

3 - is there any reason to expect GI (or PM for that matter) caustics to be messed up by interpenetrating geometry?

It shouldn't be much of a problem.

Best regards,
Vlado

[simmsimaging]

Thanks Vlado!

[kosso_olli]

Thanks Vlado.
Good to know. I think I might test a little more, as I am still missing some small light streaks araound the drops. Maybe this works with the directly visible caustics option.