Caustics: odd behaviour with different shapes

[simmsimaging]

Something else occurred to me while messing with this stuff for so many hours: it would be great to have a utility that could scan two Max files and print a list of the *differences* in Vray settings. Even if it only touched on the render dialog stuff (not materials and lights) it would be really handy for quickly figuring out how a scene has gone wrong from an earlier version, or in a case like this to rapidly see what settings are clearly different.

/b

[kosso_olli]

Could never get this to work either. We often have to deal with the same kind of jobs as simmsimaging, dealing with the same problems. In the end I just disabled the photon mapped caustics and relied on the caustics produced by the GI.
Never really understood how the whole caustics system relates to the unit setup of the max scene you are currently working in.

[simmsimaging]

Any chance to look at the problematic scene Vlado?

[MoonDoggie]

Something else occurred to me while messing with this stuff for so many hours: it would be great to have a utility that could scan two Max files and print a list of the *differences* in Vray settings. Even if it only touched on the render dialog stuff (not materials and lights) it would be really handy for quickly figuring out how a scene has gone wrong from an earlier version, or in a case like this to rapidly see what settings are clearly different.

/b

Doable. I'll see if I can find some time if some ppl want this? I agree too. Just pull the render settings from another file. It's just not useful all the time.

[kosso_olli]

Here's my try with the caustics on droplets. I had 10 different droplet shapes which were then scattered on the can using multiscatter. The caustics are precomputed using a VraySun far away from the can. I figured out that this produces much more cleaner and sharper caustics.
PS: I didn't want to have any caustics outside of the drops. As you can see, the inner part of the droplets is slightly lighter. This is from the caustics.
Like to hear your feedback.



Caustics Pass

[simmsimaging]

It looks good, but hoping still to get Vlado's input on the last files - if only to avoid adding more variables by changing scenes again

/b

[vlado]

Here are my results; mostly I just changed the caustics search radius and made them emitted from the spot light rather than the rectangle light.

Best regards,
Vlado

[simmsimaging]

Hey Vlado - thank you for looking at it.

Your result looks more or less like mine did - and I tried both lights too. The drops still look really dark over most of the sphere and it still doesn't look anything like the version from the scene derived from AJ's file, so there are still big questions to my mind:

- Which is "correct"?

- It is still not clear why there is such a huge difference when the caustic settings appear identical

- How would I adjust one to look more like the other?

Thanks

b

[vlado]

Ok, I see; the path tracing image does indeed look quite a bit different - looking into it now in more detail.

Best regards,
Vlado

[vlado]

Btw, the correct image is the one that you get with brute force GI/progressive path tracing Ideally, the photon map should produce the same caustics, but I'm looking into why they are darker.

Best regards,
Vlado

[vlado]

So this is a bug; when photon mapped caustics are enabled, V-Ray will not compute indirect light paths for the inside of the droplets and would wrongly assume that any such paths will be computed by the photon mapped caustics - so bounced light from the box below the tomato does not reach the inside of he droplets as would happen when you use regular GI caustics. I hope to fix it for next week.

Best regards,
Vlado

[simmsimaging]

Good to know there is was something amiss and I wasn't just crazy I'm curious though: not sure what you mean about the path tracing image looking different. Both scenes use photon mapped caustics, just one looks way darker - or did I miss something there in the AJ derived scene?

[kosso_olli]

Nice to hear that your are working on this, Vlado. But I am not sure about the Path Tracing version. How can we enable that to see the difference? Would like to try that.

Regards

Oliver

[kosso_olli]

Just want to bump this one: Whats the deal with the path tracing version? How can I enable that to test a few things?
Also, why is the Vray Sun a much better caustics generator than any other light?

Thanks,

Oliver

[MoonDoggie]

The Vray sun is a much better generator because it's so bright.