Cloth on Cloth Collisions

[ThE_JacO]

when garments are onion skinned the way yours are the best way is always to simplify your life.

simulate the trousers and cache the simulation, plot the animation as shapes and mute the cloth op.

then simulate the layer on top of that with cloth.

inteliigent managment of caches and ops, baking into shapes and layering simulations is your best bet for a trouble free life when handling complex simulations.

mutual deformations are most times useless and way too complicated to manage with cloth (they are actually very tricky to implement in an efficient system due to state monitoring and caching issues, only syflex 3.05's flesh deformer handles them properly in my experience).
The only instances where you could need those are usually muscle simulations and volume preserving softbodies, but those are things into a wholly different ballpark from what you are trying to do.

[NeptuneImaging]

I just did a simple test of two planes colliding with each other and the bottom plane acted like the cloth but the second planes fell toward the first on and just hit nothing from invisible planes...

Weird...

[ThE_JacO]

it's not weird, it's the annoying consequence of the state caching model that most simulation engines use to simplify calculations and increase stability (stability as in stability of the resulting spring network, not the stability of the process).

when a simulation is started the simulated body will check the list of obstacles, it will find what state they are in and cache it, then at every frame it will push that cache through the transformation matrix of the object, but this basically means that it's colliding with the mesh not deformed by its own cloth.

in collisions you can set "consider shape changes", and this will force the simulation to walk the geometry everytime instead of considering the one cached, but if both simulations run at the same level they could end up not considering each other, this is why you can still obtain the results you're after by baking the simulation into shapes for one of the meshes, because that WILL be evaluated.

mutual influence in collisions is only present in very few simulation engines accessible in XSI (syflex does consider it up to a certain extent), and even then it will very often have a one frame discrepancy depending from the order of evaluation of operators.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.