There may be an anomaly with the way AddForce with continuous force and rigidbody colliders operate, this could mean, as just one example:
If you have a rigidbody character with a jump and it uses Unity gravity, then you may get inconsistent jump heights depending on the number of colliders that the character is touching.
I recently noticed this whilst posting (and answering) an issue I had in Unity Answers:
(I won’t repost all the detail here but there’s an explanation and demo project in the answer)
I thought I’d post about this here to raise awareness (and folks can confirm or dispute the finding) and also to ask (as I don’t have access to Unity 5) if anyone knows if this issue still present in the 5 Beta?
If it isn’t fixed in Unity 5 (or if no-one replies to this :-)) I’ll log a bug. Thanks!
Yes, in my original Unity Answers post I incorrectly used Update(), but the issue still occurs with FixedUpdate() and in the Answer the V2 sample I posted also uses FixedUpdate().
I have amended the Unity Answers original post to clarify that.
I just remember that I had exactly the same issue in the past. My workaround was to multiply the force with the number of currently touching floor colliders which worked reliably as all of them had the same friction. I just made a quick test in Unity 5 and the mentioned workaround is not anymore needed.
When it comes to believable physics simulations, you always have to play with the settings and try a lot of different things out. It doesn’t matter whether you are dealing with Unity 4 or Unity 5, even though it seems to be better in Unity 5. It is always your job to change the conditions in the environment such that they become stable, which is usually not that simple.
