Jump to content
ParticleSkull

Create multiple FEM over time?

Recommended Posts

I had substep set to 1 and collision to 3 - I still think it's overlapping spheres when they are created though.

Share this post


Link to post
Share on other sites

Alright, hahaha substeps at 1 and collision to 3? now I got it!

I don't mind about the overlapping when they are created. This should be out of the screen on my final scene ;)

Share this post


Link to post
Share on other sites

the overlapping at creation creates the problem of merging spheres later on. 

  • Like 1

Share this post


Link to post
Share on other sites

That's some really useful information!

I'm still getting some spheres merging into each other btw, even using your settings. It's looking much better than what I had before though, I think I can solve it increasing the mesh resolution, collision passes or substeps.. and now the overlaping at the creation! 

Share this post


Link to post
Share on other sites

Just opened your scene and I'm pretty sure your issue is you are forcing interpenetration - in my experience, if you create intersecting meshes, the solver will allow that interpenetration from that point - and I might be completely wrong, but that's my experience.

Share this post


Link to post
Share on other sites

I needed to place slices of strawberry slices in a glass a few weeks back and did end with the same problems. Sim times were ridiculous (hours) compared to the almost realtime RBD sim (seconds). I'm also experimenting with the grain solver and actually can't tell which one is best to use for which scenario.

Since my strawberry setup looked very similar:
What I observed is that when the emitting stopped and no force anymore "pushes/hits" from above, the strawberries tended to jump out of the glass. In my feeling there's something wrong with the FEM solver or a certain parameter needs to be used/enabled. It's been like the strawberries worked like a spring. The moment the force from new strawberries stopped, the springs tended to "elongate" and expend the stored forces, jumping out of the glass. In a way storage of forces needs to be disabled, so that there is no more kinetic energy stored inside of the FEM objects. At least in cases like that.

I guess the reason for the interpenetration (I suffered from it too) is that these forces get so high, that within the sim step the force pushes the borders inside of each other, which can't be recovered afterwards. So both objects will melt together.

Any FEM/grain wizards here?!?!

  • Like 1

Share this post


Link to post
Share on other sites

Nice work, Johnny! But your scene suffers from the same problems at the end (lower left). My objects were all created on the first frame without interpenetrating each other. That CAN'T be the reason for the interpenetration later. At least not alone.

Share this post


Link to post
Share on other sites
21 minutes ago, Pancho said:

Nice work, Johnny! But your scene suffers from the same problems at the end (lower left). My objects were all created on the first frame without interpenetrating each other. That CAN'T be the reason for the interpenetration later. At least not alone.

Well, if you put enough pressure on something, on the collisions, stuff will start interpenetrating no matter what solver. These solvers are made for VFX, they balance performance and precision - thus they'll work up to a point, then stuff start to error out. As an FX TD your job is to work around it and I truly don't think this is an issue at all. Wedge 25 sims with some tweaks between them and have the client guide the process, it's how this stuff works in practice. ;)

And sure, this might be something where grain performs better, but you gotta test it out and compare. I've been messing with bullet softbody setups (like this and this) for a while to work around limitations with grains and FEM, it's all trial and error, there's no one specific solution that'll work for this or that, it all depends on what you need. If you wanted the pressure from the balls to break the flask at some point, you could do this all using bullet and constraints. :)

  • Like 1

Share this post


Link to post
Share on other sites

The point is how to avoid the "stored" force. An apple in a bowl of apples, even on the lowest level won't jump out of the bowl at any given moment. The pressure in the glass example should solely be made up out of the gravity force and the weight from the spheres ontop of it. This should never add up to any ridiculous force which make the sim go bonkers. So why do the shperes on the lowest level in the flask still penetrate each other? Seems that the structure inside of the mesh stores the force which should actually be destroyed by friction and generation of warmth by deforming the objects.

How would one do this???

 

 

Edited by Pancho

Share this post


Link to post
Share on other sites

P.S.: Is it the force which tries to reestablish the original form of the geometry? How can this force be cropped? Something like a maximum value. Even a marshmallow tries to get back in its original shape once deformed. But its force is small, so it takes a lot of time. With the spheres it should be the same.

 

Share this post


Link to post
Share on other sites

I'm digging into it now, but to be honest, I wouldn't use this solver because you don't have access to anything on SOP level (as in a SOP solver in DOPs) so my normal workflows are out the window - thus F it, not using that crap if I can avoid it. The whole point is control, imo, and seems you don't have any in regard to FEM...

And yeah, long time since I dug into FEM, but it's annoying the crap out of me, it's not a black box, but it feels like it could as well be if I need to start unlocking the subnets in DOPs and start messing inside them - something that in itself seems to make Houdini extremely crash prone... Did I say I was annoyed? Grrrrrr...

Share this post


Link to post
Share on other sites

@Farmfield, idk man, there might be something wrong with my houdini. I've downloaded your scene, put it to simulate and got a couple of interpenetrations (horrible ones!)

edit: Love the @Alpha = 0.1; thing at flask :D

Edited by ParticleSkull

Share this post


Link to post
Share on other sites

Time to submit this ticket? ;)

//SideFX Software Inc//Bug report//2017-16-09//Houdini 16.0.600//Softbody simulations doesn't calculate correctly if the software is run south of the equator//

And my personally status is of course Alpha = 1.0 :D

Share this post


Link to post
Share on other sites

hahahaha I will!

In fact the weather is crazy here. It was kinda rainy when I started the sim and sunny at the end... The temperature should have influenced it.

Serious now, the file you submitted is really working?

Btw, I've decreased the Scale Time to 0.1, increased to 30 substeps (at the dop node, it looks diferent from the fem solver one) + ABE2 and 3 collision passes and it worked!

Now i'll dig on it to see if I can improve the simulation tim (it took like 4 hours)

 

Share this post


Link to post
Share on other sites

The timestep is 0.25, the DOP substeps are 1 and the solver substeps 2 and collision substeps 2, and it works great for me. 

zhZd5Xe.gif

(the green stuff are encoding errors from imgur)

flask.v1.Odforce.FF.hiplc

Edited by Farmfield

Share this post


Link to post
Share on other sites

Naeh dude, this is all in H13.5... This one: [link]

Edit: ^the link is for 16.0.633, latest production build, I just wrote 13.5 to confuse PS a bit more - it's friday, I deserve some entertainment, right? :D

Edited by Farmfield

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×