I have been at this a few days now and cannot work out how to stop the cloth from eventually sliding over the collision geometry and producing very large polygons, i want them to stay consistant like actual cloth would stretch and hold it's form but no matter how many time steps, particle bind steps, cloth properties, geometry density i change this keeps happening, i thought this to be a fairly simple thing to do in tyflow but cloth is not acting in the correct way, is there something obvious i'm missing.
I have attached a file that reproduces it, this happens to me a lot and i'm sure i'm just doing something wrong.
08-13-2025, 04:37 PM (This post was last modified: 08-13-2025, 04:46 PM by Findus.)
Can't open your file, if you save it back to the oldest version possible, more people will be able to take a look.
From what I can tell, you are overstretching the cloth considerably. You can try to lower your stretch values, and a denser geometry and smaller timesteps should help, but this might simply be beyond what the cloth bind can deal with.
You could also try to slow down the animation. It's quite jerky, meaning big changes in a couple of frames. Scale is correct?
Also, the small regular wrinkles pattern looks like the collision radius of your Particle Physics might be too high. Or are you using CCCS? If not, try with that if you have the Pro version. Or perhaps you are using a length bias ot cause those bumps (you could mask that with a texture).
Maybe try to add a Push Out Operator to help with penetration? Use a second slightly bigger collider mesh if need be.
If this continues to cause trouble, you might be better off with faking this, e.g. via conform.
(08-13-2025, 04:37 PM)Findus Wrote: Can't open your file, if you save it back to the oldest version possible, more people will be able to take a look.
From what I can tell, you are overstretching the cloth considerably. You can try to lower your stretch values, and a denser geometry and smaller timesteps should help, but this might simply be beyond what the cloth bind can deal with.
You could also try to slow down the animation. It's quite jerky, meaning big changes in a couple of frames. Scale is correct?
Also, the small regular wrinkles pattern looks like the collision radius of your Particle Physics might be too high. Or are you using CCCS? If not, try with that if you have the Pro version. Or perhaps you are using a length bias ot cause those bumps (you could mask that with a texture).
Maybe try to add a Push Out Operator to help with penetration? Use a second slightly bigger collider mesh if need be.
If this continues to cause trouble, you might be better off with faking this, e.g. via conform.
Thanks for the reply, honestly i've tried all these things and it just doesn't seem to change much. I have added a 2023 version which is a recreation, and this is a much slower animation that shows the same issue. would value a second opinion, thanks!
08-15-2025, 10:07 AM (This post was last modified: 08-15-2025, 10:17 AM by Findus.)
(08-13-2025, 06:09 PM)ldotchopz Wrote: Thanks for the reply, honestly i've tried all these things and it just doesn't seem to change much. I have added a 2023 version which is a recreation, and this is a much slower animation that shows the same issue. would value a second opinion, thanks!
Thanks for the file! Did you recreate this to test with the slower animation or because you can't share the original file? Else you could have simply saved your file back to 2023 (under save as). Works most of the time. You probably know this, but just in case.
You are basically stretching the cloth way beyond what normal cloth would be capable of, and in my experience, the Cloth Bind works only well within very limited parameters (out of the already limited parameters that are available). That being said, the collision does seem to act funky and you'd expect for the cloth to be stretched more evenly. Here, all (over)stretching happens in the collision area and very little in the rest of the cloth. Maybe a bug, @tyFlow?
We can circumvent that by deactivating the particles in the proximity of the sphere and use object bind instead of a collision (won't work in all circumstances, but here it's fine). This way, we are forcing the cloth to stretch in between the deactivated particles.
Shortening the timesteps helps to a degree, but it also makes the cloth stiffer again.
08-15-2025, 08:46 PM (This post was last modified: 08-15-2025, 08:47 PM by ldotchopz.)
I recreated it to see if it happens in general and it does.
Thank you for your workaround, this is how i expect it work out of the box, I'm not sure this will work for what i'm doing but thank you because it's solved an issue as old as tyflow.
[/quote]
If you reduce 3 parameters (stretch, shear, bend) to 0,1, then set your "stiffness solver" to constrained, and finally set a damping to 0,3, it will work much better.
Out of all settings mentioned above, I think the most important is changing "stiffness solver" to constrained, because it seems that moving cloth influence remaining of the cloth in much better way... it moves more when dragged.
(08-16-2025, 01:05 AM)d4rk3lf Wrote: If you reduce 3 parameters (stretch, shear, bend) to 0,1, then set your "stiffness solver" to constrained, and finally set a damping to 0,3, it will work much better.
Out of all settings mentioned above, I think the most important is changing "stiffness solver" to constrained, because it seems that moving cloth influence remaining of the cloth in much better way... it moves more when dragged.
Certainly helps! thank you for attempting it. It still would not work on the lowest time step so it seems there is a lot of tweaking to get a simple thing to work. I am surprised tbh, i thought this was pretty straight forward. Appreciate the help!
Well, I am far from being knowledgeable in cloth, and this one I figured out just by tweaking settings.
It's rarely case that cloth needs to be stretched like this (even spandex would tear in reality after 1/2 distance), and probably this is why is not so straight forward.
Probably the best thing here would be to play with "modify bind" operator, because with that operator you really have control over the cloth.
Maybe it's possible to make parameter bind length stretch over time... or stretch more if it's being stretched...
We certainly need more in depth tutorials about it (how it works under the hood), and then it would be probably easy to achieve this.
08-18-2025, 08:04 AM (This post was last modified: 08-18-2025, 08:11 AM by Findus.)
Using the constrained stiffness solver is a good idea, because it "decouples" the stiffness from the time and solver steps. "Better" at rougher settings, "worse" at finer ones, but in this case "worse" is good, as it means more stretch.
Let me preface the following by saying that tyFlow is an incredible piece of software and Tyson should get some kind of lifetime achievement award for improving 3ds max so much. I'm also far from an expert in tyFlow and not a technical artist to begin with. TyFlow also offers a lot of other tools to fix things.
However, I do think cloth (being only a small part of tyFlow, which is not a dedicated cloth solution) has quite a few issues. Free flowing or single trick motion design things work fine, but for things with several constraints, e.g. actual clothing beyond skirts etc., it's a struggle and very glitchy. Better tutorials can help to a degree, but that can't solve some of the basic problems.
I understand cloth is hard to do in general and have a rough idea of why e.g. stiffer cloth is prone to twitching (even more so than normal settings), but for example in this case, yes, it's overstretched, but what happens in the collision area simply shouldn't. It should "fail" more gracefully.
