[center]
On the left is using cylinders and on the right is just using normal parts and here is the CSG physics geo for it. Looks like it doesn’t actually effect it in anyway except visually
[center]
For now you could put a touch brick right above it and when it connects, anchor + sink it.
That’s gonna be a can’t do because it’s meant to be a marble run and move with the big disc as it rotates. If I anchor it, It won’t move with the big disc 
You don’t have to use CSG physics, you know. Build a collision mesh out of invisible bricks.
Unfortunately the automated nature of the decomposition algorithm requires your details to be a larger percentage of the entire shape to be decomposed better. It has particular problems with making pockets as well.
I’m looking into a better algorithm to use, but I don’t think I’ll have any updates to it this year but I plan on working more with it Hack-Week this year. I know of a technique I can use, but unfortunately it might increase the amount of time it takes when you union parts can vary from being super-quick to super-long.
TLDR: It’s not Cylinders, its the decomp algorithm. Yes I know it’s limited when you’re trying to make details that are a small percentage of the shape’s volume, I’m researching ways to improve this but not as my primary project at the moment so no updates this year.
[quote] …
I’m looking into a better algorithm to use, but I don’t think I’ll have any updates to it this year but I plan on working more with it Hack-Week this year. I know of a technique I can use, but unfortunately it might increase the amount of time it takes when you union parts can vary from being super-quick to super-long.
TLDR: It’s not Cylinders, its the decomp algorithm. Yes I know it’s limited when you’re trying to make details that are a small percentage of the shape’s volume, I’m researching ways to improve this but not as my primary project at the moment so no updates this year. [/quote]If you do find a better algorithm during Hack-Week please, please let us try it out. Maybe finally my marble run dreams can be a reality and various other small-scaled ideas I have.
[center]
External Media
[/center]
[center]I know some people are frustrated with this as well.[/center]
I’d be willing to wait around up to 20-30 Minutes if I knew it was going to work. I’m a patient person and I don’t think time would be a concern for me.
Just turn off can-collide and make your own invisible collision boxes, like Spacek said, if collision is that important to you.
The problem is that you’re relying on a computer to do a task that could have an infinite amount of outcomes. It chooses the one it deems best, but that solution may not be the one you want.
With these algorithms there is no one-size-fits-all solution, so you’ll usually be better off just doing the work by hand. Just build the collision hull out of individual parts, that way you’re guaranteed to get what you want.
Alright I’ll resort to an collision hull [size=1]i really didnt want to a lot more work and harder[/size]
[center]I have a feeling I’m going to get a lot of this
[/center]
[center]Don’t worry, I got it to work but now it’s two onions which is going to be tricky getting it to rotate with my lil code
[/center]
while wait() do
script.Parent.CFrame = script.Parent.CFrame*CFrame.fromEulerAnglesXYZ(0,0.1,0)
end
“RIP my hopes and dreams”
bro lrn2CSG
Chop that cylinder up into a quarter piece, cut the cylinders through it, and then copy and paste it 3 more times to form a full circle