ClothBind-System produces weird artifacts after a while

[DominikW]

Hi,

I have a system with ClothBind for straw in an agricultural machine.
The reference objects are quite simple sticks that just fall above some deflectors.

Now my problem starts when I want to cache those.
It works completely fine for a while but eventually after 1000-2000 Frames it starts producing artifacts that gets bigger and bigger and then the system vanishes completely.

Is it maybe a memory issue? How can I fix this to cache till the end of my animation?

                   

[tyFlow]

Can you post the file? Hard to guess what the issue would be, especially if it only occurs after 2000 frames.

[DominikW]

Sure, I add the cleaned-up file to this post with just the particle system inside.

The time when the system starts producing errors and eventually vanish seems to depend on the number of particles which are emitted.

In this file when I hit the Generate Cash File button, the errors begin around Frame 400 and quickly get worse until complete chaos.

[tyFlow]

Thanks for the file. So I can reproduce the issue you're seeing around frame 400, but it's going to be really tough to debug considering it doesn't appear until hundreds of frames into a heavy sim, and it's likely dependent on a single numerical value somewhere in tyFlow's simulation pipeline going haywire. That type of issue is really, really tough to track down.

One thing I did notice is that if you disable all of the operators that are sending the particles to the Delete event, the bug doesn't seem to occur any more (at least, it didn't occur up until frame 500 - I didn't simulate further with them off)....so it might be related to the deletion of your cloth particles. Furthermore, I tried deactivating/stopping the particles before deleting them, and then I couldn't repro the bug even until frame 800....so maybe that's something to explore as a workaround (add a Particle Switch (deactivate mode) and a Stop operator to Event_006 before the Delete operator).

Just a general question about this setup: your timeline is 3100 frames long...is the machinery meant to move around in space at a given point in your full file or do anything that would prevent you from just setting this up as a loop? Seems like just making a 100-200 frame loop of the particles moving along that is repeated over time would be a better alternative than simulating 3k frames.

[DominikW]

Thanks for the reply, I am gonna try your workarounds and see if that helps.

Yeah, a loop might be a good idea as well. But there will always be a visible hard cut inbetween loops, right?

[tyFlow]

Not necessarily, just sim some particles falling down and disappearing (as you do), but only over a finite time range (ex: 200 frames, so at frame 150 have the emission stop so the particles all eventually disappear), then overlap the caches with an offset (ex: 100 frames). You won't notice the intersections during the overlap and there won't be a hard cut.