dForce energy issue

felis

I created a hoodie as kind of a test. It worked fairly as planned.
Then I realised, that I would get explossions, when she was posed where she was translated downwards. And after a good amount of try and error I found that it was my straps that were causing problems. The straps were attached with rigid follower nodes, but the largest part of the straps themselves could simulate.
And I found that the upper part of the straps would get a deformity, which most often led to explossion. Even if she wasn't posed but just translated downwards there would be issues, although not explossions.
So the straps generate more energy from the downwards motion, than dForce is able to distribute without issues.
I tried changing density, velocity smoothing and some dampings, but couldn't find an actual solution.
Lastly I extended timerange and number of subframes and made a succesful simulation, but even there you can see that there is kinks in the straps.
I am interested if somebody has suggestions to settings that can decrease the energy, when shifting to a lower pose, without having to extend translationtime and subframes.

Standing, all is good.

Straps at start of simulation

Straps just before explossion

Crouching pose obtained through longer timerange and dobbling of subframes

crosswind

It might be better that you set it as a Dynamic Surface Add-On as it needs an 'inner support' to avoid collaps. And may add some Velocity Smoothing to its dynamic surface to avoid distortion.

Richard Haseltine

I assume the laces are colliding with the (static) eyelets? My guess would be that there isn't really enough space there for the thickness of the laces plus twice the minimum offset (once either side) and so more enrgy keeps getign pumped in every step as the simulation tries to move them in (to allow space) and then back out to relax.

crosswind

Like this Hoodie product that has straps with 0 Dynamics Strength as default ...  Setting add-on takes time so I just reset strength on the 'Rope' surface and added a weight node. Turned off Self Collide, added more Buckling Stiffness and Velocity Smoothing. The strapes may keep its shape when draping.

felis

Setting self collision to off didn't change anything.

The first loop of polygons in the strap are set to 0 dynamics. That probably means only the first loop of vertexes are set to 0, while the second is a mix between 0 and 100, so probably 50. So I tried to expand the 0 area surface to the second loop of polygons. It didn't really change much, beside moving the problem area one row of vertexes.

I tried to set the ring to not be part of simulation. That caused Daz Studio to crash twice, as soon as I clicked 'simulate'. Then I tried to hide it with the geometry editor, and funny enough, when I clicked 'simulate' DS crashed.

Tried if air resistance and gravity did any changes, and neither did they.

Reintroduced velocity smoothing, and changed buckling stiffness. It might help, but didn't prevent explosion.

Explosions only occur when there is a downwards movement. So my interpretation is that when the strap connector (the rigid follower node) is moved down, dForce can't find a solution to distribute that movement to the rest of the strap, and the frame before explosion there is heavy kink on the first loop with dynamic strength different from 0.

So besides from doing downwards movement over a longer timeframe, I don't currently see a solution.

crosswind

Or tweak the strap's dynamic surface a bit stiffer... set Resolution Level to Base... change Collision Mode to see the result, etc. More experiment may find the way out for explosion.

BTW, it seems that you made this hoodie in Marvelous Designer. If so, making it a pair of cords may be a better option as they could be stabler for collision...

 

felis

After I have slept on it, I have some ideas for what I want to test.

And if that fails, how I will try to modify the mesh.

The hoodie is modelled in blender.

crosswind

Oh, pls do. Still, if in Blender, better make a pair of cords or drawstrings with Cylinder or Cube.

felis

I tried to simulate the strap on its own - worked fine.

Then I attached it to a static plane and simualated. Explossions when moved to much.

Then I merged it with the plane (so the mesh was attached to/part of the plane). Then simulation went fine.

All situations had the same mesh that could simulate.

I found it would be funny to have mesh attached with a rigid follower node that could simulate. I don't know how dForce handles rigid follower nodes, but there obvious is something with them. Anyway, attaching something with a rigid follower node that can simulate on its own, seems to be a dangerous road.

So in my hoodie, I extruded some new straps from the mesh, so the rigid follower nodes only holds the rings.

crosswind

Great ~ and that's almost the same effect with MD, sewed first and export as welded single obj.