Self assembly of complex object

[janeti]

I'm wondering how I might be able to show the assembly of a complex object made of interlocking parts.

For example, if I wanted to show the assembly of a soccer ball, I want to start only with black, curved pentagonal shapes and white, curved hexagonal shapes. Is it possible to use dynamics to simulate the creation of the soccer ball by creating attractive forces and some momentum to each of the white and black pieces?

I've read a little about the idea of using locators as goals and goal weights, but is there a way to make the pieces attract each other, say, only along the edges of objects (in my example, along the edge of a black pentagon or white hexagon)?

In case any of you are wondering, I'm a biologist and am interested in making a simulation of virus self-assembly. I've got Maya unlimited, and I'd rather not have to write any MEL!

Thanks!

Janet

[drblacksnake]

in soccerball case i'd say animate by hand. viruses i assume are more complex and depend on shape itself and shape of pieces You want to assembly it from.
so, if You want any suggestions, elaborate.

[janeti]

The idea is NOT to have to keyframe everything. It should be a random process, with things bumping into each other, sometimes sticking, sometimes bouncing off. This random process could eventually lead to the formation of the soccer ball, or it might not. I want to be able to change parameters, like initial velocity, curvature, number of pieces, etc., and then let it go and see what happens.

I realize that Maya may not be the best thing to use to make a simulation, that I should probably learn C++ or something, but what I want to know is whether this is even possible in Maya. Can it be done using MEL scripting?

Here is a link to a physical model of the type of thing I'd like to animate:

http://www.scripps.edu/mb/olson/movies/Tan...g-virus-web.mov

thanks,

Janet

[failure]

You can't do it without certain ammount of programming. But in general writing simulation of this process shouldn't be too hard - it's a stohastic model and you don't actually need to simulate the dynamics ( which are on virus level different than bottled demonstration anyway ), only probabilities and their results. It's doable in mel.

[drblacksnake]

umm.. since You want to have most of things random and customizable for me it sounds more like stand-alone sim or plug-in at least.
Also I guess that fellow scientists won't take You seriously, because Maya is not scientific tool, however it could be used to assist in some cases. Maya is more for visualisation of predictable events than random, since when You cannot some things achieve using "standart" methods, You can trick it around (e.g. bake simulation and correct keys manually), which is not scientific.
so, on Your question if it is possible I'd say: YES, but.. consider if it is worth it.
anyway, thats only my opinion, so listen for others.

[janeti]

Excellent advice, thanks.

Sounds like this is going to be a tough project. I heard that Houdini may be more apt for this kind of thing - anyone know whether people do these sort of simulations with Houdini?

[failure]

Houdini or Maya - it doesn't matter. Like it or not, you'll have to write some code. Making standalone program is the best solution in terms of execution efficiency but in that case you also need to write some additional code to handle the display of simulation ( which you get for free when using 3d package ).
This problem is too complex and too specific to be handled without scripting by any 3d animation app - unless, of course, you are willing to ditch the simulation approach and just animate the aproximate visualisation of the process by keyframing.

[user_deleted]

what about having it all start completely built in the final shape that you want...have all the objects start as passive rigid bodies then key them turning to active rigid bodies in whatever fashion you want, use some kind of field to blow them apart and then run the video in reverse? that will make it look like the pieces are comign together...no mel requiered...

[joojaa]

This sort of thing is usualy tough easier done in reverse. Just make a bunch of particles that movea away from their start positions in some nice random manner using fieslds expressions etc. and then render it out and play it in REVERSE.

[failure]

If I understood original question correctly the point was to have control over input parameters of the proces and to see what results different set of parameters yield.

[joojaa]

Its still easier in reverse and you get all the randomness you ever want. Just make rigd bodies and let the expression cut the constraints between them.

Tough the real problem is reatracting at edges.

[failure]

Yes but that way you are displaying deterministic model, not stohastic. Strictly speaking of course.

[janeti]

Strictly speaking, failure is right, I wanted to create a simulation, and depending on input parameters, it may not always lead to the formation of the soccer ball.

There are simpler problems that I wanted to use this sort of system for - I just put out the grandest idea since I figured if the hardest thing was possible, than easier things would definitely be.

Given that it may be overly difficult/impossible to do this simulation (without hella MEL), and I still would like to keep keyframing to a minimum, I really like the exploding-soccer-ball-played-in-reverse idea. I hadn't thought about that, and it would illustrate the idea nicely.

[joojaa]

Im not sure you understand that MEL is usualy not used in dynamic content at all but rather plain vanilla expressioons.

And exprressions is not MEL, tough beginners usualy think of expressions as mel.

In this case id say you could do it with one single line of TYPED mel. Infact its impossible for you to strikly speaking doe allmost anything without mel, since just the aact of running maya means youve uesed a hellofalot of mel.