FLIP - Mesh Improvement Tips? (stripes on viscous)

[ParticleSkull]

Guys, any tip on how to get rid of those stripes and the tiny bumps?

The simulation is pretty high res. Particle Separation is at 0.01 and i'm simulating it with 5 substeps. I think, for the stripes, it could get better with more substeps but it's already taking forever to simulate so i'd like to check if you guys could recommend me some cool approach.

Here's an image and a video showing the sim.


 

Thx

[Atom]

This is your star project right? It looks like to me it has something to do with that initial shape. How are you implementing it? Is it vector drawn or has it been rasterized converted? Maybe try modulating the shape a bit with noise after the initial reveal that establishes the shape at the start.

Looks like fun play dough though.

Edited September 29, 2016 by Atom

[ParticleSkull]

Hey Atom, yeah, it's the star shape project. But it's just a personal project trying to achieve a super high quality sim.

About the emmiter, yeap, i think you're right, I guess the problem is there. 

I'll try to increase the remesh res and add some noise on Y

I've done the star shape from a circle animating a coulpe of points so the emmiter starts as a circle and "morph" into a star. Then i've done an extrude + remesh (that could have a higher res)

Here's a video to explain it better:

 

Quote

Looks like fun play dough though.

I see what you did there - lol

Cheers,
Alvaro

[ParticleSkull]

Hey Atom, now it's working fine! The problem was on the X animation of the emitter. I was using $F to move it (noob!) so it wasn't getting the position on subframes. With $FF it's perfect.

Atough I still need to get rid of those of those bumps. A smooth surface would look pretty cool.

In case anyone else have this doubt, here's the right way:




Cheers,
/Alvaro

[Farmfield]

Test changing from the splashy to the swirly kernel, might get the surface less "bouncy", thus less bumpy.

[ParticleSkull]

Hey John,

Good idea!I'll give it a try

[Farmfield]

Hehe, I love when people take my pseudonyme first name and shortens it.

But yeah, try it out, though I'm sure you'll have to mess with the particle surface setup as well, to get it "perfect".

Edited September 30, 2016 by Farmfield

[ParticleSkull]

hahahaha, i don't even noticed that

yeah, I already got some good results by tweaking settings on the influence scale of the particlefluidsurface + a little bit of vdb smooth and adaptivity but had a feeling that some flickering would occur.

I guess swirly kernel makes a lot of sense since I don't need any splashes (not sure if this thought is right)

[Dave Stewart]

Yo Joh!  <ducks and runs>

I've found it really difficult to get initially highly-viscous sims to smooth out, even with really extreme settings, while still preserving folds/creases. Really curious if you guys figure anything else out on this.

Alvaro, you may want to ditch the shelf setup for meshing and build a custom VDB setup that lets you better mask the VDB smoothing based on curvature/laplacian analysis.

Hoping that H16 includes better meshing tools that are tailored to small-scale/viscous sims...

--Dave

[ParticleSkull]

Damn Dave, if you can't get it right I believe it might be nearly impossible. What do you think Joh? hehe

I'll do some research on how to mask using VDB smoothing based on curvature/laplacian analysis. It looks complex but very promising

Thx Dave

/Alvaro

[Farmfield]

Hilarious, I almost fell off my chair here. Joh: MTV Raps  

Anyhow, I would look into velocity smoothing, because the underlying particles creating the bumps aren't really travelling straight, but kinda wave'y, so smoothing that out, would

Or you could do the opposite, increasing the resolution even more and adding a tiny amount of positional noise and then smooth that out in the surface meshing.

Edited September 30, 2016 by Farmfield