So right now newton physics and cloth don't really play nice together. I've been fiddling around a tiny bit to see how this could be "faked" to some degree. Here's my plan when I get a chance, or maybe some of you want to try it out.

This could be as simple or complex as needed. The idea is using small newton objects (static with action). These objects will have path constraints to the cloth mesh. See where this is going? These "proxy" cloth thingies will interact with other newton objects. Where it gets complicated is how many would you need? In a perfect world you would want one for each grid in the cloth. That will be my starting point for a test, a small cloth grid and bunches of small models path constrained to the cloth mesh.

So "static with action" would work with the cloth effecting newton objects; keeping things in a cloth bag for instance. Each object would be assigned a cloth deflector which would cause the cloth to move and also the cloth newton objects to effect the other newton objects. It gets a bit fuzzy in my head till I test it out.

What about newton effecting the cloth? Trickier I think. So now we put bones on the cloth. The bones are constrained to the newton cloth "proxies" (instead of path constraints). The newton cloth thingies will now interact freely with other newton objects. How will the cloth behave? Will it be able to act on it's own normally if nothing is hitting it?

Tests to come.

If I am way behind on this... if you guys know of something I don't like please let me know.

-vern

I don't think this will work after all. Which is simulated first? Cloth or physics?

-vern

Well you are in control of that, arent you?(You pick) I would imagine, off the cuff- that you would want to run your Newton first...and with that all simmed then you would do the cloth so you would get natural 2ndary motion off your Newton objects... I don't know your particulars tho-

The chicken! No, the egg! It sounds like they are pretty interdependent. I don't really know much about the NewtonPhysics plug-in, but here's an idea.

If you have a bag full of stuff, model the bag and simulate the stuff in the bag. Then, assuming your bag has the opening at the top, simulate the cloth with gravity going upwards, so it contours to the stuff in the bag. Make sense?

[ETA: this approach might not work if you are trying to do something like set the bag of stuff on a table. But thinking about it, maybe it would turn out okay. I guess you have to try.]

I submitted a feature request on AM Reports. I don't expect miracles.

Here's my idea:

First off newton is limited in what can be a dynamic object and the mesh animation. For instance a static object with action can have a deforming mesh over time. From my testing a dynamic object can not have an animated mesh. It changes the center point and the outer "surface". For instance a deforming cloth mesh absolutely can not be used as a dynamic object because the outer bounds of the mesh are constantly changing. It can be used as static with action but this does not give good results.

My proposed solution is in two parts. First deal with this limitation in newton and cloth by using the concept outlined in my first post, dynamic objects constrained at the intersections of a cloth grid. I did this with an action. Two constraints for each path adjusting the ease so the line up at the intersection. These dynamic objects follow the cloth and act as cloth "proxies". It "worked" but of course it didn't work really because both sims need to happen together.

The second part would be the simultaneous simulation of both systems. It wouldn't be that critical in which order it happens as long as each system takes into account changes that occur from the other system. Sort of like "taking turns" back and forth. The cloth moves the constrained dynamic objects which in turn would effect other dynamic objects. Those other dynamic objects would have a cloth deflector material, moving the cloth which once again, moves the constrained dynamic cloth "proxies" at the intersections.

During both types of simulations each frame is calculated before moving to the next and starting again. So the trick is to sim cloth, then sim physics, move to the next frame and repeat.

This is not a perfect solution but completely avoids changing the behavior of either system. They are exactly the same but simulate "at the same time" but not really.

-vern