Help I can't get dForce to explode
Padone
Posts: 3,481
Yes, I'm trying to do a test where it explodes.
It's common believe that dForce explodes when the cloth is pinched between two colliding objects. For example when the hand touches any part of the body that is covered with clothing. So I did an animation doing exactly that. Two simple spheres colliding on a simple cloth. And it doesn't explode.
I also tried to start the animation with the cloth colliding inside the sphere. And it doesn't explode either. It just stays pinched in the sphere and goes on fine.
So what's in the dForce Outfits that explodes the geometry ? Can we reduce it to simple tests ? And why my examples don't explode ?
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I find using the wind effector can make a lot of outfits explode.
It could be the base mesh is not high-poly enough in your tests. Try it with clothing that has more polygons against bodies with more polygons.
And your "common belief" comment makes it sound like conjecture and speculation. I've seen dForce explode clothing. I've run over a dozen simulations trying to find the correct settings for clothing that was not made expressly for dForce, and I've seen these mesh explosions with my own eyes. I'm just not polluting YouTube with videos of each one, largely because that would be stupid, and because that's not the result I was going for.
"Common belief". Pfft. Try "FACT", son.
Hilarious. You're joking, right ? :)
What kind of animation are you doing ? It's hard to tell from your screenshot, is it a simple y-axis movement, sort of a sphere drop ?
Generally it's a sheering force and geometry intersection that causes "explosion" . Velocity is part of the equation , if motion is very slow, or up the iterations, you can reduce the chance of explosion.
There are many workarounds such as tweaking settings, friction, changing the animation, using invisible colliders , etc... but dforce is easily the most unstable cloth simulation engine out there. It's a very unforgiving engine. But hey, it's free and it's an early beta in development, it should get better
Yes, other cloth engines can explode under similar circumstances too, but less frequently in my experience. They tend to be more stable, more mature, more features - but they too had many problems early in their beta phases too
dforce is the "king" for exploding progressive new-age artwork scenes. :)
@pdr0
Tnank you so much. That was exactly the kind of example I was looking for. And I agree that dForce is very explosion prone. That's why I'm trying to figure it out with simple examples. Because it can help to understand better what's to be avoided in the setup/animation. My example was a vertical motion. The below sphere moving up to the upper sphere.
I use Blender for production. That, among other things, I believe has a much better simulation system. So my tests with dForce are just for fun.
@DrNewcenstain
I had no intention to mean that dForce explosions are not real and/or just speculations. I guess that's just my bad english. Though I check with the google traslator before posting, I fear some expressions I choose may not sound completely fair.
I was just trying to figure it out with simple examples to better understand what's the cause and how to avoid it.
That's a great animation; really shows the problem clearly. If there was more accurate collision detection along with appropriate deformation of the objects then the moving sphere wouldn't be able to "grab" the cloth. I guess we will just have to wait for that level of accuracy in the collision/deformation algorithms.
And, just for the sake of completeness, this is how Blender handles the @pdr0 test. No explosions either. It elegantly first collides the cloth with the first sphere, then pick it up with the second sphere. I didn't specify anything fancy, just default values for the sym with one plane and two colliding spheres.
Finally I found some time to do more tests with dForce. What I found is that I can greatly reduce the risk of explosions by using the "good-discrete" collision mode and reducing friction to the minimum. Below the example with the standard G8F starter pack and a hand penetrating the clothing. It doesn't explode.
I tweaked some parameters in the simulation but it's just to make it faster since I don't like waiting. Scene included for anyone interested.
Also I found that I can get a behaviour similar to Blender by using different collision layers for the three objects. This is the @pdr0 test with dForce and no explosions. I used the same sym settings as in my previous example. Scene included.
And I can also decide if the cloth gets grabbed or not by changing the friction values. As always no explosions. This seems pretty flexible overall.
I have to admit that dForce made to me a very bad first impression. Even with the provided "dForce ready" content in the G8F pack. With the DAZ default settings it was just exploding for whatever minimal reason and it was awfully slow either. But now I'm beginning to understand that with some tweaking it can do quite fine.
Here I used the same "trick" to get G8F grab her gown. I just assigned different layers and changed the friction in the Arms surface. I guess it would have been worked the same by using an invisible grabbing sphere parented to the hand. It doesn't explode.
Beware that in this case if you get the hand too far from the body then it does explode. But this would be the case when a real dress would rip apart. So I feel an explosion may be quite acceptable.
EDIT. I believe it would be useful to have an animable friction value. So that you could change it during the animation to get a "grab and leave" effect. Having an animable on-off value would do fine just the same. I don't know if it's already possible in some way. Given the limited animation features in DAZ Studio I feel not.
And in the process I also learned how to create and upload animated gifs that can be much more explanatory in these cases. So thanks to @pdr0 for pushing me in that direction.
And finally this is my first attempt at doing some dynamic hair with dForce and the G8F pack. Since I'm lazy and it's just a test I simply applied a gradient falloff with the weight editor. But it turns out it works quite good as simple as it is. Scene included.
This is a test with the Charm Hair included in G2F Day at the Beach Pack. I did this just to test dForce with some long hair. As always it's just a gradient falloff with self collision off, so it's a very simple setup. The result is not optimal though it's not that horrible either. I believe the lack of "realism" in the animation is due to the low number of strands in the hair. Scene included.
Those transmapped textured ribbons will always look quite rubbish
and will never pass for actual hair in an animation IMHO
Perhaps for the one framers ..but not in animation.
Yes I agree. At least they will not pass as realistic hair. I believe having proper shaded fibermesh hair would deliver better results. I'm doing these tests just for fun and to get out what can be done with dForce and actual DAZ assets.
Having hair or anything else clipping with the body in zero pose before sim reliably causes explosion for me. It's often a problem with setting up hairs for dForce because they have unseen bits clipping the skull in back and no smoothing modifier by default (adding one sometimes fixes it).
What I do is to use the good-discrete collision mode and reduce friction to the minimum value. For hair I also disable self-collision. And I always assign different collision layers for best control. This seems to greatly reduce the risk of explosions.
Of course I agree that "unclipping" geometry before simulation is always a good thing as well. But in my tests above with the explained settigs, clipped geometry just tends to stay sticked and it doesn't explode. Unless there's something stretching it very far where a "real" cloth would rip apart.
But it also seems to depend on the "complexity" of the mesh itself .. so no real silver bullets here.
Maybe Daz can strike a deal to get Nvidia Hairworks into DS. It'd be nice.
I already pointed this out in another discussion so this is here just for completeness.
Below are some dForce settings you can use to speedup simulation. Be aware that lowering subframes and iterations will make the cloth "sloppy", but it may be fine anyway depending on the scene you're doing. Or you can compensate by reducing gravity and/or the dynamic strenght.