large world floating precision

So, I was thinking how to go about solving this, and found this video: https://www.youtube.com/watch?v=VKWvAuTGVrQ. And this article : http://floatingorigin.com/mirror/continuous-world.htm

Now, I was thinking more along the lines of what Unity has done in that video, just basically moving the entire scene, rather than have each piece have its own coordinates.

However, wouldn't doing this break the static colliders? Or can you move an entire scene by root and not effect the colliders? Otherwise, every single static collider will need a kinematic rigidbody, to help the performance. But even then... thats a lot of rigidbodies. Or maybe a hack way, would be to throw a kinematic rigidbody on the scene root object, but that seems like it would be... disastrous.

So, my question is simply that, can you move static colliders by root without breaking the performance of the static colliders? Or would kinematic rigid bodies need to be applied?

In contrast, how can we set up a per chunk coordinate system like gas powered games has, in Unity? I found this question with something similar:

http://answers.unity3d.com/question...-coordinate-system-changing-spaceworld-1.html

- but as I posted, how can we accurately determine a chunk origin if the chunk itself is far off? Seems like this wouldn't work either.

 

Any thoughts? I'm really interested on the effect putting a kinematic ridigbody on the entire scene would have.