Corona 1.4 & Caustic

[nicolasZ]

Hi guys,

I'm tring to render pool with caustic but it does works !
I read some theards about that and if I understand well experimental features like VCM or BIDIR are not any more on the 1.4 version.

So if you know how to achive this effect, please tell me how to do that.

Many thanks !!!

[TomG]

For a pool, my thought would be to fake it by projecting the caustic pattern using a light - as an example in 3ds Max, add a spotlight, set to rectangular, open the Advanced Effects tab, drop a texture that represents the colours of swimming pool caustics (should be plenty of free ones out there on the web) into the Projector Map box, then to make setting it up easy I'd ensure the walls and floor of the pool are separate objects and use an Include / Exclude list under General Parameters so that the light only affected those.

Hope this helps!

[PROH]

Hi. Theres a couple of threads in the forum about "fake" caustic, but if you're searching for textures, then here's a link to 8 animated sets of caustic:
http://www.dgp.toronto.edu/people/stam/reality/Research/PeriodicCaustics/index.html

[cecofuli]

Unfortunately, Corona in Progressive isn't able to render Caustics. Only with PPM (Corona v1.4 - Experimental rollout)
But, PPM doesn't support A LOOOOT features of regular Corona (for example, Absorption). In fact, it's only an experimental engine.
We must to wait Ondra and a better caustics optimization for Progressive engine =(
You can render with PPM and mask the pool.

PS: these are real caustics, not a fake. And they are very fast!

[romullus]

I read some theards about that and if I understand well experimental features like VCM or BIDIR are not any more on the 1.4 version.

Those experimental engines wasn't removed, but hidden from casual users. You can find them in devel/debug rollout. But first you have to unhide debug rollout itself, you can do that in performance tab.

Also you may want to read some user guides about various approach at caustics:

https://forum.corona-renderer.com/index.php/topic,3093.0.html
https://forum.corona-renderer.com/index.php/topic,5431.0.html
https://forum.corona-renderer.com/index.php/topic,10343.0.html
https://forum.corona-renderer.com/index.php/topic,10779.0.html
https://forum.corona-renderer.com/index.php/topic,11630.0.html

[cecofuli]

Yes romullus, but in the v1.4, we haven't this rollout.



And, if you select Bidir/VMC in the v1.4, I think that it's set as [PT+MIT] and you cannot create caustics with Bidir anymore, right now.
But, with a sting, I think that you can select  [VCM] (good for pool caustics)

In conclusion, with the v1.4, you can generate caustics only with PPM.

[PROH]

Hi cecofuli. That's some really nice caustic you made there (PPM). I had to try that myself, but I can't get anything as nice as yours - not even close. So I wondered if you would share some tricks or maybe even a sample file?

Regards

[cecofuli]

It's real 3D water, with closed volume, a lot of subdivision and a little bit bigger than the pool,
The top part = noise modifier.
The shader, you must to check  "Caustics (slow)"
I cannot send you the file, sorry.

[PROH]

Thanks! I used displaced water, might be that.... I'll try with noise mod.

Thanks again.

[Frood]

but in the v1.4, we haven't this rollout.

It´s still there. Have another look.

Good Luck

[cecofuli]

Are you sure? O_O

[Frood]

1.4 Build timestamp: May 16 2016 11:07:17 (final). Or did you mean 1.5?

Good Luck

[cecofuli]

Honestly, I never saw this rollout, after A6...  O_O I'm shocked... how can be possible? =)

[Frood]

how can be possible? =)

Maybe you have never scrolled since A6 :)

Good Luck!

[cecofuli]

No no... as you can see with my screenshot, our rollout title is the same, but mine is gray-out...
Does anyone have the same bahevior?

[Frood]

Err... don´t you just have to press "+" in the rollout title bar? In your screenshot it´s collapsed.

BTW: Yours is not 1.4 final.

Good Luck!

[cecofuli]

(O_____O)'

[Cheesemsmsm]

Why are you still using daily build?
mine is the same as Frood's

[Juraj]

PS: these are real caustics, not a fake. And they are very fast!

Nope, they are not fast. I tried using this on few occasions, only the water, to composite it with rest of the image with progressive in post.

Even though they docaustics, and they become visible rather instantly, the noise would never clear. It would do the funny 70 000 passses (I kept it running for 24 hours just for fun) and there would still be noise.

They're not advisable to use because they don't support feature, but because they don't even work well imho. Which is sad...because Caustics would fucking rock !

[Ondra]

the thing about photons is that they can sometimes never converge, and even if they do, they converge asymptotically slower than path tracing - that means that as the render progresses, they become infinitely slower than path tracing. So juraj's observed behavior is consistent with the theory

[Juraj]

Although I wonder what denoising would do, but this combination probably doesn't work together.

[Ondra]

it would probably work pretty bad, since its theory is derived for uncorrelated monte carlo noise, not for correlated photon noise. Also "sample" has completely different meaning in photon mapping than in path tracing

[gerets]

More test and attemps...this is my result :

[cecofuli]

Juraj_Talcik, I mean, they are fast than Progressive or any other method in Corona.
Obviously, we cannot compare to V-Ray caustics or with any other Renderer.  But... in some way, they works =)

[Ondra]

More test and attemps...this is my result :

Looks nice, how exactly did you render that?

[romullus]

More test and attemps...this is my result :

Looks nice, how exactly did you render that?

This isn't Corona, look at those glowing balls - no shadow terminator :] Just kidding, author please share the knowledge, i think many reaaaly would like to know how to do such beautiful pools :]

[PROH]

Totally agrees :) I would love to know how to do such nice pools!

[gerets]

just moment...1-2 hours and I can share my experience with some image

[cecofuli]

I think that it's possible with Corona+VCM & Compositing the pool with the rest of the image.
You can have a nice Caustics and good GI.

[Juraj]

Does VCM support absorbtion ?

[cecofuli]

No, only Progressive support Absorption.

[Juraj]

So how can you composite a nice pool if the VCM will produce pool with caustics, but without absorbtion ? The water will not look that good with just one of these.

I tried overlaying just caustics using screen, but even this is not good solution since caustics cause illumination that I otherwise not have without them (non-caustic refraction looks different, not extremely but it's not it).

[cecofuli]

In the second image, I don't see any absorption...
In the first, the absorption looks like "painted by hand" =)

[gerets]

1) render with Progressive (water with absorption)
2) render with Bidirectional with simplified scene pool (water caustics)
3) compositing

[gerets]

obviously:



PS render VCM is fast (1 hour)

[gerets]

two simple tips but very important:

1) color diffuse material wall and floor pool similar at absorption color water material. This is important for compositing

2) simplified scene for caustic calcolation. This is important for render time with Bidirectional PPM

simple, fast and effective method




and anyway...love also no caustic water

[cecofuli]

Thanks Gerets for the tips.

[Mr.Schorsch]

These Images have by far one of the best greenery Setups i have ever seen. Such love in placing all to flowers and plant and Patels. Stunning work. And thanks for telling us the caustic Setup you used.

[tomislavn]

Your vegetation and details are absolutely amazing!! I am loving those images - it's such stunning work, really good job! Oh, and thank you for the tips :)

[Giona]

This is a really interesting thread!
I tried to create some caustics, but with my scene I was able to achieve a good results with different engine settings. Instead of PPM, I used Vertex Connection.

Here is the result:

[cecofuli]

Try to render the pool with a white shader.
 Now, it's difficult to see if there is something wrong.

[nicolasZ]

Wooo, so much interesting things !
Thanks  for the tips Gerets, and your images are awesome.

But I can't manage to reproduce the effect. I don't understand what I'm doing wrong ....
For the water did you used a plane with bump ? And what kinf of lighting did you used ?

Here are my settings, if you could tell me what is the problem, that would be great !

Thanks
 

[cecofuli]

Here the max file.

[nicolasZ]

You rock ! Thank a lot cecofuli !

[Giona]

Here's the test with a white material inside the pool.

I also tried another scene to see reflective caustics on the walls. I tested the scene with BDPM and Vertex connection, 4 min each scene and the results are almost the same.
It could be super usefull to have a render pass to save only caustics..

[burnin]

Very nice results.

Does it work in C4D version? Is engine on pair?

@giona4
What do you use for edges (modeling, shader...)?
Looking good.

Here's simple scene test, rendered with BD/VCM (BD-PM option is not available), caustics for material on, lighting sun+sky only, water surface displaced. Resulted in fireflies and much darker refraction than w/o caustics. Should report a bug?

Edit/Note:
Done with Corona for Cinema 4D Alpha version: A6 daily Jun  2 2016 (core 1.4 DailyBuild Feb 11 2016) - reported on Mantis (ID 1984)

[FrostKiwi]

I ran a caustic test scene and looked at how the Rays are being distributed.
The SamplingFocus render element is normalized to 0-255 in the attachment.

Caustics hitting the Floor had on average a realRGB of ~1.25 with the highest couple being ~2.
Caustics as seen through a refraction had on average a realRGB of ~3.5 with the highest couple being 7.

Adaptivity samples caustics through refraction more than twice as often as not seen through a refraction.
While it is correct to assume, that adding refraction makes the chance of a Ray spazzing out a random color and creating noise, it is the exact opposite in this case.
The most noise was produced from the caustic rays on the floor being scattered across the Plane. They created the most visible noise, (partly due to the fact, that the floor had a solid color and thus makes noise more visible) yet still was sampled only half as often, as caustics as seen through a refraction, which did not contribute to the Subjective Noise as much.

Kinda like the Mantis post I made, where glossy reflection of a glossy reflection was sampled twice as much (which didn't produce much noise) vs the first bounce of GI, (which produced the most amount of noise), yet was sampled only a quarter as much as the average of the whole image.

edit:
Scene was rendered in Progressive with 1250 passes and 0 MSI.

[Radim Razzak]

Hello everybody,

This might be a really stupid question, but how can I enable this Experimental section in the Performance tab? It looks like it's hidden from the casual users too well. :)

Thank you

[TomG]

Hidden with reason :)

In 1.5, you can find it by enabling the Devel/Debug mode - go to System tab, System Settings button, check Enable devel/debug mode, close the menu and head on over to the Performance tab.

[Radim Razzak]

Hidden with reason :)

In 1.5, you can find it by enabling the Devel/Debug mode - go to System tab, System Settings button, check Enable devel/debug mode, close the menu and head on over to the Performance tab.

Great, thank you very much!

[Radim Razzak]

Well, after having followed the hints and tricks mentioned above, I managed to get some caustics in my test render. But the bad news is that I get these white artifacts even after 3 hours of rendering on my dual xeon. Any idea on how I could get rid of these or what might be causing them? I just used a simple HDRI setup for fhe lighting.

Thank you.

[FrostKiwi]

Well, after having followed the hints and tricks mentioned above, I managed to get some caustics in my test render. But the bad news is that I get these white artifacts even after 3 hours of rendering on my dual xeon. Any idea on how I could get rid of these or what might be causing them? I just used a simple HDRI setup for fhe lighting.

Thank you.

No easy solution this time. It's an experimental render engine, that is ultra naive.
Sampling into a pixels works by averaging the cast sample with the previous result. A reflective ray with a >255 energy will take days to average out against the only 128 blue. These are extreme fireflies.
This is usually solved by blocking rays from having high energy with the MSI option. Problem is, that BiDir doesn't honor that option and Caustics are by definition high energy rays and thus you can't really use that option.

So you have to do it manually in post. You may be lucky to be able to just crush that information inside the Corona VFB, by looking at the intensity of the firefly with right click and setting Highlight compression to roughly the roughly same amount + some more to get it under white.

But most likely that will look aweful, use Corona's built in firefly filter option.
or
Export as any HDR format.
Either:
Go into Photoshop and try Median Filter with 1px or 2px amount, or
Use any firefly filter like ArionFX's Despeckle, or
Use Burntool to paint the Fireflies to 0-255 range by hand.

edit:
Remove Outliers filter in ImageJ worked wonderfully.
(I'm sure you can find many other tools, currently programming with it, so had it on my desktop :D)
Free tool, can download it here.

[Radim Razzak]

Thank you very much SairesArt!