Earlier I've animated simple collision with a cloth-modifier. (I've compressed it to a 2MB mp4, but I don't know how to upload it). It shows a face moving along a path through a (pinned) cloth, so that the contours of the face come through. Like this:

enter image description here

Currently I'm trying to achieve the same effect in different scene, but now the cloth seems to 'explode' even before the face hits it. For some reason the cloth seems to diverge into infinity, even though all the settings (at least in the cloth-physics tab) are the same as in my previous animation.

enter image description here

My question is: why? What causes this behavior?

I've tried to remove and add the cloth serveral times, but I keep getting the same results. I've checked:

  • The normals (are pointing all in the same direction);
  • Remove doubles (or 'merge by distance' as it's now called);
  • Apply all transforms to the cloth;
  • Increase the quality steps both in the cloth- and collisionstab;
  • Increase and decrease the subsurface levels of the clothobject;
  • Check and uncheck 'self-collision'.

I get the impression that the behavior has something to do with the last two, for sometimes it works okay (though undetailed) with subsurface level 2, but not with level 3. Similar with the 'self-collision' check. The behavior isn't consistent though.

Finally, here is a screenshot of the cloth-settings:

enter image description here

And one of the (subsurface) modifier(s):

enter image description here

I hope anyone can help me to understand this behavior.

  • $\begingroup$ What are the dimensions of your cloth in scene units? Cloth simulations tend to get glitchy and unstable at small scale factors like 0.1 units and below. It is best to scale everything by a factor of 10 run the simulation then scale back down if all you need are static frames. $\endgroup$ – Duarte Farrajota Ramos Oct 4 '19 at 11:10
  • $\begingroup$ The dimensions of my cloth are X: 1.725, Y: 0, and Z: 0.898. I'm trying to figure out how to scale up everything with the same factor, while keeping all the proportional distances between the objects the same. When I have, I'll let you know whether it solves the issue. $\endgroup$ – Mathew Oct 4 '19 at 16:11
  • $\begingroup$ @Duarte Farrajota Ramos You were right! I scaled everything up with a factor 5 and that seems to resolve the strange 'diverging' behavior. I hope it stays this way on subsequent simulations, but it seems promising. Curious that the scale can have such an impact. // The remaining problem is that with the exact same cloth-settings the current cloth-simulation turns out much coarser than the one I'm trying to remake, but perhaps that's just because I've changed the cloth from a square to a more rectangular shape. $\endgroup$ – Mathew Oct 5 '19 at 19:20
  • $\begingroup$ FCOL, after adding a new cache and adjusting the mass of the cloth, it starts all diverging again :( The cloth dimensions are now all above 5, so the scale cannot be the problem anymore. $\endgroup$ – Mathew Oct 5 '19 at 20:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.