Seems in Blender that there are 2 ways to attach an armature to a mesh: the “old way” being parenting the mesh to the armature and the “new way” being adding an Armature Modifier to the mesh. There seems to be confusion over the benefits of using the new way. I have been confused not so much that the new way is cleaner and suits the overarching workflow of Blender better but with all the facets that rigging encompasses (weight painting, vertex grouping etc). For instance eyeballs as a separate object could be parented to a bone whereas, as far as I can see, if you apply the Armature Modifier to the eyes you then have to link the vertices to the bones – many more steps.
The “new feature” of Bone Heat Weighting has only confused me more as the workflow for that seems to include using the “old method”. So if they are shifting paradigms why are there new (and attractive looking) features for the old method.
Links
http://dreamsgate.wordpress.com/2008/06/09/rigging-bone-heat-weighting/
http://www.blender.org/development/release-logs/blender-246/skinning/
http://wiki.blender.org/index.php/BSoD/Introduction_to_Rigging/Preface:_Rigging_in_Blender
As shown on this page the “new method” is “better” which is somewhat unhelpful especially when viewing the “new method”s feature page (http://wiki.blender.org/index.php/Manual/PartII/Modelling/Modifier/Armatures) which doesn’t really list the aforementioned features that make it better than parenting the mesh. Another thing that I still don’t understand from the instructions here is why one would need to “give the mesh Copy Location, Copy Rotation and Copy Scale constraints, all targeting the armature” – isn’t the modifier supposed to take care of that?