Hi all,

Please bear with me if this is something basic, I'm just getting started.

I'm trying to create a simple, floppy, dynamic fish, but I want to be able to animate its tail as well (so it can flop around once on the floor). I assumed the best and/or easiest method for doing this was by animating the rest geometry of the cloth object, but I can't get it to work. I assume the rest geometry is only evaluated at the first simulation frame?

I tried the squishy object route, but that doesn't look remotely right . I'd rather not go the route of building the fish out of smaller RBD blocks with links, if that will work at all. Seems an elastic object approach is the easiest to get looking right, untill you try and keep the poor fish alive.

Any ideas?
thanks,
naam

Why not use deforming RBDs along with keyframe animated flapping fish?

Have a look at the attached example file. Just enable Deforming Geometry on the RBD Object DOP.

Also had to set Resolve Penetrations to 1.

Thanks! Much appreciated, but not exactly what I'm looking for. I want the fish-object to react like a soft body, so the force of a fall could bend its tail, maybe squash a little bit, but still have the fish itself now and then twitch or bend. The problem with using an RBD is that when the fish is motionless, it will look stiff as a plank.

Do you know if it's simply impossible to use an animated rest state?

I'm going to see if the same can be done by using animated springs to act like muscles maybe. Or maybe use a SOP solver? No experience there…

Thanks again!
naam

Well I tried and tried to get the desired effect by using a few animated SBD springs pulling at points on the mesh to simulate the twitching, but I'm getting nowhere. The results look promising initially, but the cloth solver doesn't like what is happening one bit, and keeps on failing due to ‘numerical problems’ every now and then, and taking way too long at other times. Kind of an unstable solution.

So I ask again: anyone know if you can make the cloth solver accept a changing rest state? If I do a simulation with a straight fish mesh, but feed it a reststate of a bent fish, it it looks exactly right, but since the reststate seems to get stored at the start of the simulation, the poor fish will remain bent. Which looks rather silly.

thanks,
naam

How about a jiggle chop?

Naam,

can't you post (part of) your file, it will be easier for us to tinker with and come up with a solution. Also, do you need to simulate the fall of the fish? Why not animate it?

When it hits the ground maybe you could get the effect you want with assigning different masses/weight to groups of the body.

Cheers,
Hans

Ok, here goes then. 3 files attached, and an obj for the fish base model.

1 - The gummy approach comes closest to what I'm envisioning, but as you can see the mesh gets all screwed up now and then. Nevertheless, it reacts nicely to the twitching, jumping up in a somewhat natural manner.

2 - The reststate approach is more of an illustrative base file, where I tried to make the cloth solver react to a changing reststate. Impossible, it seems, but I'd love to be proven wrong there .

3 - The springs approach holds some promise – I'm ‘squeezing’ two places on the fish's body together there, hoping that the body will bend naturally. But this takes a lot of tweaking and before you know it the simulation gives up. Also, you have to use two spring constraints, one to counteract the other so the fish doesn't fly off into space . In all, a method that will probably require way too much tweaking to be feasible (imagine having 8 of these fish interacting .

There's some crud from failed experiments in this last file too, please ignore .

The reason I'm trying the simulation approach at all is because I'm investigating in how far Houdini wil be able to help us with an upcoming project – which as you might imagine isn't about a single falling fish – and learn how Houdini works in the process. There's plenty of other methods, like animated RBD's, or even just particles combined with hand-animation, but I'm secretly hoping to give the thing some much-needed ‘definition’ by going for a full-on simulation approach.

Oh and thanks Soothsayer, but as far as I can tell you're just animating the mesh after the sim? The sim doesn't react to the mesh being warped, right?

thanks,
naam

Ah great some study material for the weekend! I was almost afraid i had some free time to do fun things like movies and pubs…