Create a part and an attachment with a LinearVelocity force. Set the VectorVelocity property to a vector that doesn’t follow the main axis or the 45°, example: (3, 0, 9), (4, 0, 7). Toggle on and off the Enabled property and notice the inaccurate acceleration motion.
As of this moment, the only work around would be to switch all the LinearVelocity instances to VectorForces, however this is not viable in our time budget.
Issue Area: Engine Issue Type: Other Impact: High Frequency: Constantly Date First Experienced: 2022-10-06 00:10:00 (-04:00) Date Last Experienced: 2022-10-06 00:10:00 (-04:00)
I didn’t even know that LinearVelocity existed until reading this post - I just used a BodyVelocity to accomplish the same thing. So if they’re not working, that could be a good alternative for now!
This isn’t necessarily true. In your case, your part is also fighting against friction, which can be enough to throw a LinearVelocity off track if it has a low enough MaxForce.
In your place file, I changed MaxForce from 500 to 5000, and it no longer went off track.
Additionally, you have the Velocity applied in the attachment’s space. So, if the part rotates at all, it will move in a different direction.
I understand this is a viable solution in this reproduction file, however this isn’t viable in my current situation. Increasing MaxForce has other consequences, such as making my object fling when colliding with other objects in the environment. More MaxForce definitely resolves this issue, but the point of MaxForce is not to fix irregular force application.
I originally discovered this issue with a floating sphere that didn’t have any ground friction (or any other surface friction whatsoever). I took the time to make another reproduction file to make the situation a bit more accurate. The part floats steadily with a VectorForce, the LinearVelocity is now applied to World, there’s also a Rigid AlignOrientation.
Please point out if I’m doing anything wrong, I’d love for this to be my own mistake.
I’m surprised I forgot the existence of the body movers. I’ve just tried them and it seems they’re probably using the same back-end as LinearVelocity now because the issue is exactly the same.
