too much energy in a fractured dynamic simulation

[Brian Jones]

you’ve read Dynamics/Tips and Tricks/The party’s Over? You’re playing with the correct settings but sometimes it’s necessary to keyframe the Angular/Linear Damping (and the other settings) to very strong values when you want it to stop, or as it says ‘steamroller tactics’ and set Time Scale (Project/Dynamics/General) to 0% – that kills all dynamics though…

[Steve Bentley]

I haven’t read that but I will. Thanks Brian.
Yes I have been keyframing those settings as well as the velocity thresholds (I figured they would have to kill it for sure but it can be tough to know what the actual velocity is of x object is) and if I take those values too high I get things hanging impossibly middair.

I thought going more accurate with the mesh of the object would help so any false intersections wouldn’t cause the sudden eruptions of movement. I’ve found that any crazy flyaway object are usually when two objects suddenly find themselves intersecting on one frame – long past just touching, due to approximations in the shape choice or by the auto shape choice changing what it thinks will work best. I figured more accurate mesh math would solve that, and it did, but introduced a huge amount of “tim robbins jacobs ladder” kind of shaking, which the step settings (now up around 75) wouldn’t solve.

I think part of the problem might be the impact object. It was part of another spot the client wants to match but it got a cache on it at some point (I don’t rememeber baking it) and now if I remove that cache in the dynamics tag, try as I might, I can’t get that dynamic driven movement again. I’m actually surprised it works at all: a cached object’s motion intersecting with objects that have dynamic tags. The big issues are always the objects that get trapped between the impact object and other static areas. If the impact object didn’t have a predetermined caches action on it, it might stop sooner and theyby not force other shapes into spaces they can’t really fit, where they reverberate off a rock and hard place trying to get out, and they do that forever, or suddenly find a way out and emerge at escape velocities.

[Steve Bentley]

I didn’t realize that “the party’s over” was in the manual, so yes I had in fact read that eons ago.
The big problem is that the manual is just worded so badly – whenever I go back to it I end up yelling at the screen. I suppose once you know the techniques the verbiage makes sense sometimes but I feel for people coming up the chain who are new to the controls or concepts. I wonder if the German one is better?

I haven’t had to work with the Restitution Lifetime setting yet but I’m starting to think I need to. Now if only the text in the manual made any sense. There are so many examples of things that are just wrong in the help pages, I never know whether to believe something that seems a little off or if it is indeed just plain wrong.

Everything written about this particular control seems to make sense..until.
It seems to imply that there are little standoff bumps that are calculated so that objects that are touching aren’t touching as far as the dynamics are concerned, so you can have (in the example) a newtons cradle at rest. So far so good. And once that value (which has no units!!!!) is elapsed – so I’m guessing either frames or seconds – then those virtual bumps go away, and now even dynamics sees them touching and will act to push them apart if bounce is set above zero (again i’m assuming because it’s not explicit).
But then it says “if the time set is too short the cradle will not work at all”. I would think if the time was set too short it would work all too well and the balls would spring apart and then start clacking. I’m hoping they meant that the cradle will not work correctly and the initial bounce will be even before any ball is set in motion, but again… manuals can never be too explicit!

Any way, that’s my rant. If I had the time I’d offer to rewrite it (“be the change you want to see in the world”, but they turned the mic off before gandhi had time to say – “if you have the time”)

[Brian Jones]

I just remembered I didn’t see you refer to Collision Noise in your initial thread, pardon if I missed it or you just didn’t put it in but that should definitely be set to 0% for stuff like this. It adds energy to make things more believably random but it keeps adding a small bit of energy at each collision if it’s not turned off…

[Steve Bentley]

Yep, already looked after – I have a feeling that “collision noise” is part of the restitution setting.

I think sometimes you get lucky with object placement and simulations and things go well and sometimes… well this is the latter.
I love their comment in the manual about even thunderstorms affecting things in sims. Its so true.
I think if I could release the cache for the impact object things would be better because then I could cache the sim. But caching just one of the new tags seems to kill the cache on the impact object’s tag so it doesn’t take much to chaznge things that I’ve gotten looking good. and I just can’t get the same motion from the impact object if I remove it’s cache.

The other day we were wondering if there was a bit of mass based gravity in each object – just like the real world – if you were in freefall. We added a new large object with a static dynamics tag to a scene and everything started falling towards it. Take it away and we’re back to normal. I would have thought only a gravity force or the projects gravity would be able to affect direction (there is only a gravity force, no project gravity and collisions in this sim, and yet we were getting a non “down” vector for our falling motion with the big object hanging around. Einstein was right- even in C4d!