OTOY Forums

radiance wrote:there's a tutorial by philbo somewhere (can't remember) on the forum.
maybe use the search function, it includes a diagram of how to make the liquids/meshes.

Handily, it's very easy to modify the technique I used and arrive at the one Philbo describes. Back shortly with results!

I'll be posting in more depth on this later, but here are the results of the initial re-render.

Couple'a changes here. First, I matched the camera settings in the earlier Maxwell Render image, so now the Octane render and the Maxwell renders can be compared more directly. I still can't quite match the materials, especially the ground texture, but that's not essential in evaluating the glass here.

I don't have the patience to do the logical step and do a photograph and 3d models to compare. Though I did take a picture of a vaguely similar glass with vaguely similar liquid.

To my eye, the Octane render isn't quite right. I don't see the interreflection between the glass and liquid layer where the glass is thickest. It makes the glass look like it has uniform thickness. Strangely, I can see the boundary at the bottom, but the outside looks 'filled' with liquid too. I'm still not sure why the base appears black. It isn't intersecting.

hey,

one thing to understand is that you won't get the same image currently as octane uses simple transmission colours instead of absorption with distance using beer's law like maxwell does.
this is a feature i'd like to add once the volumetric foundation is done, which will be for v2.0 probably.

Radiance

It's right here:

http://www.refractivesoftware.com/forum ... f=23&t=649

try not to set the roughness of the ground too low, or you'll get a lot of fireflies... and the refraction flipped the background upside down? what do you think of it?

radiance wrote: one thing to understand is that you won't get the same image currently as octane uses simple transmission colours instead of absorption with distance using beer's law like maxwell does.
this is a feature i'd like to add once the volumetric foundation is done, which will be for v2.0 probably.

Yeah, I realized that pretty quickly -- but my workflow is extremely messy.

I did go back and render out another Maxwell shot using much less absorption, but really, it's kind of beside the point. What I actually want to be matching is the internal reflection between liquid and glass and there are better ways to get that right. PhilBo's technique is still not looking quite right to my eye. I suspect it's a modeling error on my part, since the liquid and the meniscus surfaces are open and don't quite touch.

Proupin wrote:try not to set the roughness of the ground too low, or you'll get a lot of fireflies... and the refraction flipped the background upside down? what do you think of it?

Yeah, that's not looking so very right to me... But I think it's because my model isn't water tight. So to speak!

What i have hear is that the should slightly intersect.
Just enough to get the right effect, it has worked for me so itæs worth a try.

Further experiments...

My time lately has been split between writing (my ostensible career at the moment), sculpting in ZBrush and playing with Octane. Octane has gotten the short end of the stick for a few days while I was doing other things, but now I'm back and working on the liquid-in-glass thing.

I'm happy to report A Discovery.

If you take a look at the attached picture (about an hour of render time), you'll see that the glass on the left is Very Messed Up. It uses simple overlapping geometry. A glass volume, completely sealed and a similarly sealed liquid volume. That doesn't work. The other two are two very simple variations. The glass in the middle uses the exact technique PhilBo described. It works nicely. The glass on the far right had a 1mm gap between the open fluid model and the single-plane meniscus/surface model. Not really any appreciable difference between the gapped and ungapped model.

However, setting this up, I kept getting annoying artifacts in the glass. The only way to make it render correctly was to flip the polygons on the liquid model (so the normals face inward).

So, to be clear, when using PhilBo's glass + liquid technique, it is important that the surface normals of the liquid (but not the meniscus/top layer) face inward.

The full glass image hasn't sufficiently resolved to be certain yet, but it looks like this trick fixed the problem I was having at the beginning of this thread too. I'll post an update later, when it's done rendering.

The new renders is interesting, i did one test myself with the meshes intersecting and didnt get any artifacts...
So i dont know what the problem is, and i have so many diffrent versions about this that i dont know wich one is true either

But please keep on testing