In my opinion, if you want to guarantee it works on all objects you use it on, that’d probably be the best method, and bounding boxes are just cubes so they have few vertices to cover if you had to. Alternatively if all of your objects are of a similar size/shape, you could get away with using the bounds size normalized as a fuzzy “zoom” variable but that’d be more error prone on large or oblong objects. Are you going to be doing this rapidly on tons of objects at once or just one at a time like a map editor scenario?