[Bf-committers] some more improvements for CA

[Gustavo Muñoz]

Thank you, Drew, I was who wrote that. I was not in this list at the time, and then I forget to post it here, where it belongs for sure. And I can now add some more related points.

I think I now understand better the core of the current NLA issues in Blender: open a new project, add a mesh, exit edit mode and move it some. Now duplicate the object, and then move it some more. Now
advance some frames and you will see you have the first nonsense: both have linked transformations, even if you have not activated the "duplicate with object: Ipo" option. In fact, both have now the same
linked Ipo, when I have not animated them. I didn´t want any Ipo...

Another incarnation of the same nonsense: If you look at the NLA window, you can see also that you have now one keyframe added for every object you have just moved at frame 1. Say you are working in a
character animation and you need to change some scenery models at the middle of your character work. You cannot just grab the object and move it, because you will animate it (remember the keys at the NLA and
Ipo windows), so you must remember to go to frame1 every time you need to correct some unanimated model... not very usefull. I know you can have allways an NLA or Ipo window opened and you can go and delete
those extra keys everytime, everywhere, but this could be greatly improved while at the same time we get Blender´s NLA capabilities increased with some changes.

Another case of the same problem is that once you begin to build an action for an Armature, you can see that your keyframing is also being linked by default to the NLA main timeline. But you have not asked
for this. If I want an Ipo for an object, I will ask for it. But in most cases I don´t want it by default. If I want an action to be used in the NLA main timeline, I will put it there. Furthermore: once you
have your character animated in an action, which is currently by default editable in the NLA window, you no longer have access to the default configuration of his meshes or Armatures. If you need to edit
them after being animated the character, the original pose pops up from the "limbo" and apears in the middle of everywhere, and without any relation to his own Armature. .. Also bad.

Sorry for the long notes, but I wanted to explain very well the importance of the changes I am asking for. And here it goes my proposition:
We need a default state for every model, where we have the meshes, the objects and the Armatures at their default poses and conditions. It must not be linked to any timeline, in order to be editable at any
moment without distractions, and furthermore for being able to change it while we see the model updated in the middle of an animation with our editions. I have used untill now a dedicated action called
"default" to get quickly this needed "default state", but this is clearly not enough: once modelled and rigged a character, and after animating it, very often you need to re-edit here and there the meshes or
the bones to get a better behaviour in some deformations. If we could get coded this "default state", we could allways open it up at the same time that we see the character in the middle of an animation, ane
edit the meshes and armatures while looking at the effects in real time.

This "default state" must not be related to any timeline, acting like a first level over which you can add and mix several animations. Every animation must be totally independent of the others, untill you
put it in a main timeline (NLA window), where you can blend them as you like. At the same time, absolutly every animation should happen in an action, an independent animation block. Vertex animations too, it
is very needed. Then we need a way to link any two or more actions (armature actions, vertex actions, etc) so that you can control one from the other. That is how Animation:Master does his "Smartskins", a
very powerfull character animation tool: You create a new smartskin (a special kind of vertex action linked to a bone´s rotations) for a bone in an Armature and you get a new independent twindow where you
rotate the target bone and then you edit the mesh to get a nice deformation at this point. Then you rotate some more the bone and repeat you edits for another angle of the bone. When you have done all the
needed deformations, you just exit the smartskin mode and it is stored with the model so that whenever you rotate this bone in a bones action, it will aply the recorded vertex animation to get allways nice
deformations. I have heard XSI have something similar, but haven´t had access. Compared to this tool, all the deformers and cluster of Maya are just not so usefull and powerfull. You can allways use
"setdriven keys" for these purposes, but it is much more slow.

Well, I am sorry for the really long mesage, but I think it is a very important matter for the future development of Blender´s animation capabilities. That are my two cents for it... ;)

Best regards,
Gustavo Muñoz