Are bones actually elongated, or is that just a visual aid

Hello everyone,

Many 3d animation programs show bones as elongated "objects". Such that when you rotate the bone, it seems to pivot around some point. I recently started working on some active ragdoll stuff, and am now manually setting up colliders among the bones for that. Of course, i could set up a separate hierarchy of colliders on an entirely different game object for the ragdoll, but I would like to try something different instead. I've reached the point where i have to decide whether to move and/or scale the colliders, so that they "match" with each other, leaving no gaps between them. But since the colliders are in exactly the same position as the bones they will reference rotations from, i'm hesitant whether moving the colliders is a wise decision. I can make my conclusion once someone can confirm whether "bones" are just uniformly scaled (x:1, y:1, z:1) structures, or if they are actually scaled as depicted in Blender, Maya, UMotion, etc.

Thank you for your time.

 

The elongated shape of bones is used during the skinning authoring process, but other than that it's just visual aid. A bone in unity is just a point in space, vertices bound to it (skinning) and a placement in hierarchy.

You're aware there's a ragdoll wizard in Unity that will automatically set up colliders and constraints?

 

Already answered but I'll say same thing in different way :

Yes bone is only visual aid. There is not actually a bone, only the joints. And a joint is really just a way of organizing things. Rather than moving many vertices, we assign many vertices to a single joint (weighting), and move the joint instead. But at the bottom level these are all just points in 3d space.

The bones represent the hierarchy the joints are in. Elbow bone connects to the shoulder bone, shoulder bone connects to the... you know the rest.

 

A bone is a visual aid.

A bone is a point. With translation+rotation+scale matrix associated with it. TO make it easier work with, it is visualized as a stick.

I think moving colliders might not be a good idea, as when collider does not have a rigidbody associated with it, the game consideres it to be a static collider, and updating it is slower.

Some games, however, support "stretch hierarchy nodes" where a bone can stretch without scale propagating to children. In this case bone is a stick, but even in this case it only uses its original length to calculate stretch factor.

 

One time that the bone's visuals could matter(in Blender for example) is with automated vertex weights. There is a feature that uses "envelopes" which have sizes based on the visual size of the bones(which is based on dragging the tail around). Then the envelopes check which vertices are inside them and adds the weights accordingly in a nice automatic manner. This is a nice way to get vertex weights done faster. It often needs some tweaking afterwards to get it perfect but supplies a good starting point.

 

Great, i know now what to do.