tyFlow Forum
2d cloth solver very very slow - Printable Version

+- tyFlow Forum (https://forum.tyflow.com)
+-- Forum: tyFlow Discussion (https://forum.tyflow.com/forum-1.html)
+--- Forum: Bugs and Issues (https://forum.tyflow.com/forum-3.html)
+--- Thread: 2d cloth solver very very slow (/thread-4262.html)



2d cloth solver very very slow - steve3d - 05-18-2024

Hi there,

I'm trying to re-create the following effect - this tutorial is using C4D. 

https://www.youtube.com/watch?v=yIiIdxRvgYg&ab_channel=SketchyVisuals.

I was told I need to use the 2d cloth solver to get similar results but the problem is that even with one piece of cloth it is extremely slow to compute when switch to 2d solver. 

It is taking 14 seconds to computer 1 frame. If the 2d solver is off the time to compute is pretty much instant. Surely 14 seconds per frame can't be right, especially when compared to the tutorial where the simulation is even interactive whilst it runs...and super quick. Is this a limitation of tyflow?

I have a very basic setup...

I have an AMD threadripper, 256gb RAM, running windows 10 64bit, 4090 GPU. Tyflow Version: 1.103


RE: 2d cloth solver very very slow - steve3d - 05-18-2024

So to do 19 frames took over 10 mins.

I then tried switching off CUDA and the result was pretty much instant. Problem is now though that I'm getting cloth mesh intersection issues. I turned on CUDA again and turned off self collisions which is what seems to be what is causing the massive computational times.

I obviously need self collisions so not sure what to do from here? This type of solution seems unusable right now.


RE: 2d cloth solver very very slow - tyFlow - 05-20-2024

The tyFlow official CCCS example files come with a 2D spline solver example similar to your reference: tyFlow_cloth_CCCS_splineSoup_001.max.


RE: 2d cloth solver very very slow - steve3d - 05-20-2024

(05-20-2024, 04:09 AM)tyFlow Wrote: The tyFlow official CCCS example files come with a 2D spline solver example similar to your reference: tyFlow_cloth_CCCS_splineSoup_001.max.

Thanks. I think the comparison I'm drawing though is that C4D cloth simulation does this with cloth and not 2d splines. I'm just wondering why tyflow cannot replicate exactly the setup of the C4D cloth, ie. Using a mesh cloth with height rather than splines...and with a much faster simulation it seems.

The other thing is that the tyflow results don't seem to produce those lovely deep curves that the C4D cloth does. It feels like you can't control the cloth quite as well in Tyflow. Saying that, maybe that's just my lack of knowledge of how to use Tyflow Smile


RE: 2d cloth solver very very slow - tyFlow - 05-20-2024

There are no 3D collisions happening in your reference video, so there's no reason to simulate as cloth instead of splines. If you want you can I guess, but it'll just unnecessarily inflate the simulation time. As splines it will solve very fast. Then you just extrude the splines for an identical look.

You can use the Angle Bind operator to enforce stiffer bend angles.

See attached.


RE: 2d cloth solver very very slow - steve3d - 05-23-2024

Thanks. I'll check out this file.

I'm curious though, is there a reason why the C4D cloth solver for this type of animation is interactive and almost real-time whereas Tyflow seems slow in comparison? Obviously I love Tyflow but just curious as to why such a large difference.