using vertex groups and performance

[landeTLS]

Recently iwe been experimenting with merged objects controlled using vertex groups in blender for my rigged/animated parts. Mainly in the hope that a single object containing several meshes would produce less drawcalls and therefore increase performance especcially in parts that have really many moving parts. (Ie rather than having over 30 seperate objects that move in relation to eachother in an animation just have one object containing all the meshes and where each part is controlled by a bone via vertex groups setup for mechanic rigging. Now in doing this ordinary object to bone parenting goes out the door and partmodules(other than ones that just play an animation ofcourse) would not(?) be able to move those the individual meshes in the object so this can only be used for stuff like the folding out animation for solarpanels etc but there has to be seperate objects where the pivot point is. Or for landing legs where the only seperated objects are where the springing accurs and the foot.

Ofcourse transforms and colliders also need to be seperate objects.

Im really a noob in modelling and on how unity handles model calls so id love some info on if this is actually a good thing im doing in regards to performance? What im noticing tho is that a bonus to this is that my scene looks a lot more organized and uv unwrapping is a lot faster also mirroring the rig or making arrays is very simple. There is no difference in the amount of tris in the part from this as none of the meshes are changed when merging. Tho setting up the vertex groups can be a bit labour intensive and also needs a few extra bones.

Edited March 21, 2014 by landeTLS