Indigo Renderer Forum

yonosoy wrote:Without bubbles then? I´ll wait to confirm a re-run of the first version.
Oscar, could you give me the direction of the facebook page above?

The normals of the bubble geometries should be reversed in order to work, did you do that ?

I'm enjoying your thread yonosoy !

Hi CTZ'n. If I understand correctly the trick is to inverse the normals and to give the same water material to them.
Yesterday I did a test and has the same look. I believe bubbles are correct. In the scene, air with higher precedence than water was used.
Hello Zom-B, I forgot to say this to you. So many thanks too.
An underwater test scene. Basically, a square kilometer pool of 600 meters of depth.
Starting with it ...

Hi Ono.
Finally the mesh resolution in "caustics over white" is fixed. The thing is that the ocean surface simulator of houdini has a resolution parameter too. Now is well balanced an looks very interesting. I will let render it till a good convergence point. Bubbles are out (they are correct but don´t has a good visualization impact).
Happy if you like to include it into the gallery. Technical gallery, I suppose ...
Thanks anyway. Maybe 2 days of process. See you all, thanks.

On the bubbles, sorry to bother you but I realize now that they should have an air specular medium, not a water material as you said. You need to create an air material for bubbles.

Hi CTZn. It is a trick!
The point is that inside the reversed bubbles indigo consider the universal air medium.
This silly question could help a lot more over to represent foam, because with thousands or millions of bubbles this system is seriously faster. Just to avoid the contact with the water surface and we see ...

That shark looks really promising...

Ending up with this.
Zom-B is right about normals (was limited to 80º, but still a little odd in the top).
Oscar is right about boubbles (the shadow inside of them are not trully). For this an arch glass material was used.
A little tip for MLT. For this lasts images problems with convergence was founded raising the value of large probably of mutation. Another focus could be decrease the value for maximum rejections (with a value of 500 instead of the preseted value of 1000). No fireflies and the best convergence possible with the preseted values of 0.4 and 0.01. This one has 4.5k samples pixel instead of the one of 8k of the last day.

Oscar J wrote:I would say that those bubbles need to look a little more "bubbly"

Yeah At the moment they seem more like a creepy skin abnormality

no offense to the render which is spot on, obviously, just joking mate

That last one looks really nice.
I was really curious what prevents the in my opinion realistic behavior of the air bubbles. So i also did a test. Sorry for ripping off your scene. I only will post it here

Seems like its just a really really hard thing to render.
You can see caustics begin to catch on at the intersection of the bubbles and the bottle. But they would need a really ling time to clean up. So I suspect these kind of SSDSS (Specular, Specular, Diffuse, Specular, Specular) pathways are not something the render takes lightly.

kklors rendered millions of bubbles in his pool render a few posts down in this thread: http://www.indigorenderer.com/forum/vie ... 0&start=75

He's rendering without MLT, maybe something worth trying?

Hi Silverwings.
Thanks for your curiosity, your time and your explanations.
This is a shot with path tracing bidir. Surprised. Some caustics are expected...
But the bubbles look realistic. Now I`m able to recognize they...
So the resolution for this pic will rest till the 4 version is out and I`ll give it a chance with photon mapping.
Raphael, can you explain the procedence of the meshes (air-water) in your test? Seems a little simulation. I like very much the behaviour of the water surface at the top (is only a mesh?).
Thanks for all.

Oscar, thanks again. Are you sure that kklors use PT for this pool? Maybe I have a strange condition in the scene...

Well he said so in his thread. And he used an exit portal for the pool as well, I think.

Yeah, I say let it rest until we have photon mapping.

yonosoy wrote:Hi Silverwings.
Thanks for your curiosity, your time and your explanations.
This is a shot with path tracing bidir. Surprised. Some caustics are expected...
But the bubbles look realistic. Now I`m able to recognize they...
So the resolution for this pic will rest till the 4 version is out and I`ll give it a chance with photon mapping.
Raphael, can you explain the procedence of the meshes (air-water) in your test? Seems a little simulation. I like very much the behaviour of the water surface at the top (is only a mesh?).
Thanks for all.

The Water is just a cube surrounding the scene. The camera is inside.
The Cube has a Water material (Specular IOR 1.33) with some SSS and Absorption to get it a little bluish and a little foggy.

Attached to the air bubbles is a nother specular material this time with IOR 1 (for air) that also has a higher precedence then the water.

I also tried combining the mesh of the bubbles with that of the water and inverting the normals of the bubbles to the inside. It gives the same result then with a air material and a higher precedence.

The "waves" of the surface is just a mesh deformer and a highly subdivided mesh. I did not simulate anything.

I hope that answers the questions

Cheers,
Raphael

Hi

The "waves" of the surface is just a mesh deformer and a highly subdivided mesh. I did not simulate anything.

This surface is resting on the top side of the cube (water)?
Render time?

Thank you

contegufo wrote:Hi

The "waves" of the surface is just a mesh deformer and a highly subdivided mesh. I did not simulate anything.

This surface is resting on the top side of the cube (water)?
Render time?

Thank you

The cube is textured with a specular material that has a bit of SSS.
The top side of the cube is deformed. Thats it

I don´t remember the render time. But I guess it was 20-30 Minues on my 3 Machine setup.

Cheers,
Raphael