Is there any possibility of methods of Frost cut off unnecessary parts of the Mesh? In the case of Water droplets on the bottle it is necessary to cut part of the submerged in the glass part of the mesh.
I wouldn’t go with cutting, but with flattening.
Here is an example:
*I created a Sphere with 64 segments.
*Distributed 200 particles on the surface using PFlow
*Turned the particles to a Frost mesh with Metaballs mode, Radius 5, Absolute Spacing 0.5
*Added a Vol.Select modifier, set to Vertex level, Mesh Object and picked the Sphere as the mesh
*Added a Push modifier on top and set the Push Value to -5 - this pushes the vertices inside the sphere along their normals back to the center of the droplet by the negative Frost radius
*Added a Relax modifier with Value 1.0 and 100 Iterations affecting only the selection to smooth out the pushed vertices
*Added a Mesh Select on top to stop the vertex selection from flowing up
*Added a Relax modifier on top with Value 1.0 and 10 Interations to relax the complete Frost
This might not be so easy when working with more complex shapes where the Vol.Select is not precise enough to select the exact vertices. But I tested with a Teapot, too and it worked pretty well.
You could experiment with different values and modifiers to get the exact result you need, also you could offset the particles from the surface a bit so the droplet centers are not exactly on the surface and you would require less pushing and relaxing…
Hope this helps.
FROST_WaterDropletsOnSphere_v001.zip (28 KB)
Hm… seems like solution, thank you.
I would like to see a decision on the level of mesh creation (may be in the new versions)
A way to cut off regions of the Frost mesh is on our wish list for a future release. For the time being you’ll need to use a workaround, like the one Bobo suggested.
Has there been any progress on this? I’m looking at another job now with water droplets on glass and this would be massively helpful.
tks
/b
Sorry, not yet.
Having spent a lot more time playing with this kind of thing, I am really loving Frost. It’s pretty amazing what it can do.
Still, I would love to see a built-in option for Frost to either clip geometry via another object, or perhaps to streamline Bobo’s method internally if it could be made to be more accurate that way.
Intersection is okay when working in Vray, but it won’t work for Maxwell or any other unbiased engine (that I know of anyway, or Octane etc), however for those the compression/Bobo method isn’t precise enough to create a good match in surfaces with finer/complex geometry and really small fine Frost shapes. The end result at current is that it still won’t render okay in Maxwell, for example, due to even tiny intersections and air-gaps between the drops and the surface. For that reason a boolean approach is really necessary IMO, but if the vertex select/compression thing could be done with precision it would be okay too.
Another thing for me: while Frost is the best I’ve seen so far for generating meshes to suit random/odd sized and shaped water droplets. However, typically the more odd/random shapes tend to lie flatter on surfaces (the more rounded ones bead up more.) So the sort of droplets I think Frost is potentially the best for, it can’t really create perfectly well because the shape ranges needed. It would be perfect if we could adjust the height/radius of the drops in the direction of the source surface normals of arbitrary geo, to stretch them into shapes that are flatter and hug a surface better. Anisotropic gets closer for sure, but it’s not quite the same.
The same thing could help replace the Vertex selection approach to compress the internal geo in the Bobo method outlined in this thread so it would be good for some non-boolean requirement projects too.
So ideally we could create a shape whose radius is adjustable/stretchable in one direction relative to the surface normal of an arbitrary shape (like a pflow source) and can be booleaned by that shape. The icing on the cake would be if it could also create a bit of curl at the contact edge where the boolean happens so it looks totally realistic (the surface tension effect at the point where the liquid meets the surface) but that last is probably too much to hope for 
Anyway, I know this is on your wish list in one form or another, and Frost already rocks, but wanted to share my thoughts on what could help for a more “ideal” type meshing for this kind of liquid/surface interaction.
/b
Just an example of the shapes I have in mind here: the majority of these should lie a lot flatter to look more realistic. (btw: shoving them further into the red ball could create something similar to what is needed, but is problematic for a range of reasons. It’s not super viable anyway.)
I think I have come up with a way to improve on this result a lot - will post something a bit later if it works out okay.
Here’s where I’m at:
It’s definitely getting a lot more realistic. I still find it a bit… puffy, but much closer. I did it this time by using a slightly more complex pflow (complex in my world, not yours 
). It’s setup with spawning particles fanning out from the original ones for a few frames. With a much higher particle count, but in clusters, I could use a much smaller radius value in the Frost mesh.
I’m using a Relax modifier to get the drops to flatten out a bit, but either haven’t found a good setting yet, or that modifier just isn’t the tool for the job.
The only other issue I’m having is getting the Pflow to distribute the particles more closely packed. For some reason even with high birth rates I always get a lot of space. Separation is on for the birth object, but set quite low. Any lower and I get these clusters of colliding particles with the same large gaps between them.
Leads me to another thing I tried: I used Forest Pack Pro to distribute clusters of spheres to use that as a base for Frost. That way I can control the positions and volume of particles very easily, and get the count way higher than Pflow without major lag. Unfortunately Frost won’t read it as a source object. Is there any way that could be done or is it not really possible?
(meshing the FPP puts me in the same laggy territory as Pflow)
/b