[Aniki]

Aniki, what you do is OK, except I don't think you should worry about the direction of normals. It should render the same look even with the normals facing opposite direction.

Are you sure ? I meant the liquid body, cause if I didnt switch the normals it would have mixed normals, on the glass part and the liquid part, maxwell doesnt care ?

[simmsimaging]

Aniki: thanks for posting the pics.

The way you have it is how I did it in the past (more or less) but I thought that way was just a hack to get around the interface problem, I figured there would be a different canonical approach.

The Thomas An derived method is not "wrong" and has worked well in the past for different setups for sure

http://maxwellrender.com/forum/viewtopi ... tersection

/b

[Aniki]

well you certainly cannot apply that method for your pouring approach, thats what I meant as wrong way

hope the images helped you !

cheers

Aniki

[simmsimaging]

well you certainly cannot apply that method for your pouring approach, thats what I meant as wrong way

Aniki

That much was clear Thanks for the input.
b

[tom]

Aniki, what you do is OK, except I don't think you should worry about the direction of normals. It should render the same look even with the normals facing opposite direction.

Are you sure ? I meant the liquid body, cause if I didnt switch the normals it would have mixed normals, on the glass part and the liquid part, maxwell doesnt care ?

Yes, it doesn't care where your normals are pointing to. Except emitters.

[Bubbaloo]

But it does care about coplanar faces.

[Bubbaloo]

I think if you are doing "liquid in glass" pouring animations, the render will be incorrect, physically. For now, your choices are:

Coplanar faces = bad
Small gap between the glass and liquid = not good
Small intersection between the glass and liquid = acceptable?

Unless someone can come up with a way to apply the Thomas An. method to many frames of dense irregular meshes, I think it'll be a limitation of Maxwell.

[simmsimaging]

The easiest answer by far is the intersection method, but it seems to totally kill caustics and that pretty much kills one of the major attractions for me to use Maxwell for this kind of stuff.

/b

[Bubbaloo]

A solution is definitely needed. Maybe a "treat coplanar faces as a single normal", but then you'd still have to boolean the high res meshes for every frame. How about a "show liquid in glass correctly" button.

[simmsimaging]

A solution is definitely needed. Maybe a "treat coplanar faces as a single normal", but then you'd still have to boolean the high res meshes for every frame. How about a "show liquid in glass correctly" button.

I like it. Maybe they could put it right beside the "perfect render" button

I really like the method used by Vray (and modo, forgot about that). It's super simple - but even that has drawbacks. It still has oddities and inconsistencies with stuff like GI caustics and photon mapped caustics for interpenetrating objects. I'm sure it's a non-trivial issue, but it sure would be nice to have a working answer for it that didn't require crash-happy boolean operations.

/b

[Bubbaloo]

My water mesh penetrated into the glass for this animation, and caustics were produced nicely. But a still render that will be studied and analyzed wouldn't hold up.

http://vimeo.com/15920676

[simmsimaging]

Animations are more forgiving than a 6-8K still, but that works pretty well for sure

I do think the caustics are affected though: you can see caustics through the glass portion, but they seem to disappear as the liquid fills it. I can definitely see a difference here with water drops that are a hair above a surface and ones that are a hair into it. It's like turning off the lights.

/b

[simmsimaging]

Aniki: did you have any problems with faceting in your glass after the boolean cut was done? I'm getting major smoothing problems in mine, just wondering if that's a Max thing or a Maxwell thing...? Hard to fix because Turbosmoothing doesn't work well on the booleaned mesh so not sure how to get around it. The Smooth modifier doesn't seem to help.

Thanks /b

[Aniki]

Not really, only on the edges it can result in slight facing problems on single polys. Yet I have two modes of boole and switching between one of those works 99 percent of the time

[simmsimaging]

Ok, thanks. My modeler is looking into the C4D approach, so hopefully that will resolve it. As long as he can figure it out then I'm good

b