cloth collision dosnt work

Hello everybody!

I’ve got a problem with the cloth simulator. In my scene theres a curtain which has to collide with a curtain rod. Even if I activate the “Collision” on the curtain rod the curtain is falling throw the rod. I don’t know what am I doing wrong.

I’d followed some tutorials but in my case something went wrong.

Has someone an idea what I’m doing wrong? I use Blender version 2.74.

Thanks a lot to anybody who’s trying to help me!

Best regards!

Hi Martin,

I made a quick test and it works for me
Without seeing your blend file it is difficult to suggest anything
Maybe the collision on the cloth object?

Best of luck


Post script …

If the cloth ‘falls through’ the rod slowly then it could be the collision distance for the cloth
The default is very small {0.015 as above}
Increase the cloth collision distance {0.05} and add a repel force {1.0} and distance {0.1}

Better if you share the .blend

Here is my blend file

What version of Blender use these tutorials that you’ve seen?
From version 2.74 there have been many changes with cloth simulation:

There are some reports I mention in this thread. I find it really very difficult to work with Cloth simulation on newer versions of Blender, it is very slow and I do not understand it. You try to use version “2.73a” which is the last version where I was able to work well with cloth:

You use “2.73a”, not 2.73. 2.73 has a bug where cloth collision does not work at all.

Short description of error
Cloth simulation run at 15 fps in 2.74 RC2 but with 30 fps in 2.73 and with 60 in 2.72a.
Bake the sim is also much slower in 2.74 RC2

Is this still true? I was going to play around with cloth sim this week, so I would do best to do the sim in 2.72a (or does that too have the bug in 2.73?) and then bake the result and export to 2.75? So confusing…how can this take so long to resolve, and how could it happen in the first place?

Seems Genscher gave up on Blender development, last commit log a year ago :frowning:

I do not know, I thought there might be a regression in the code so that at least return to the last known state where it worked well.