As part of class, the professor decided for us to post in here in order to give us some points. Its an introductory class about rigging, so as an animator, what kind of interface do you prefer? controllers, selection systems, shelves and stuff? thank u!
I’ve found that animators tend to prefer a more tactile interface, and to that effect they really enjoy moving controls around a figure. That’s not to say that there are not those who prefer a GUI. It really just depends on what the animator wants in a particular situation. For reels’ sake however unless you’re showcasing a fantastic new feature, GUIs tend to look more impressive.
Hmmm… well if the professor “decided” this for you then I guess we’re obliged to help. 
I find most animators prefer whatever selection tools or controls they’ve used most recently in the past. However, I agree with Sean, as a rule controls that are visceral are appreciated more than a separate selection gui. Floating window guis are more useful in cinematic rigging, facial rigging, or for use with any rig that has many, many controls that are hard to click.
I personally dislike Maya rigs that use control curves as selection objects. I know that this is the norm, I just find that I mis-click a lot. I always liked how Max rigs have a volume that you use to select, so when I rig in Maya I try to emulate that. I use semi-transparent nurbs surfaces for controls. I have a slider that controls how transparent the shaders on the nurbs surfaces are, and as I get used to the rig I usually slide the controls to be transparent. This lets me focus more on the character. When I select a control I still get the nurbs selection lines, so I see what I am working with. But when I deselect the control the rig disappears. Then I can scrub the timeline and just look at the motion directly on my character.
I’ve attached an image to demonstrate the above. This was a rig I did a while back for a friend’s side project.
As for “selection systems” I used to just set up whatever my animators asked for. At one studio they used a series of selection sets that are nested in a similar fashion to the control hierarchy. So, for example, there would be a selection set for all of the finger controls, which was under a selection set for the arm controls, which was under a selection set for the body controls. I found that this was a pain to maintain as rigs change and new controls are added or removed. These days I use one selection shelf for all of my projects. Once the initial shelf is written it is very quick to set up a rig to use this. I’ve attached an image of what the shelf looks like as well. The premiss works like this:
I already have these control surfaces with a shader on them. I create a unique shader for each limb, for each dangly thing or tail, for the body, and for the head. Then I assign that shader to the controls that it should influence. For example, the l_leg_material is assigned to every control that makes up the left leg. If you select the Left Leg button from the shelf this something similar to this script runs:
ctrlShapes = cmds.hyperShade(objects= ‘l_leg_material’)
The actual script is really different, but this gives you the basic idea. I select every control that is connected to that shader, and as a bonus I have a list of all the controls in the left leg control set. This way even if I add 20 controls to the left leg, or change the names of controls on the left leg, or whatever madness happens… as long as I assign the shader to the proper controls the selection group will continue to work. I make the control shape shaders different colors so that I can visually see if I forget to add the correct one. This strategy has the added bonus of giving you a very clear place to put meta-data. Any meta-data needed for my ik/fk switch or mirror tools sits on the shader for the appropriate limb.
I went on for a bit, but I’m not 100% sure what you are looking for. If you have a more specific question, an example of something that you are thinking of, or a more specific region of your character that you are looking for advice on then I think you’d get more responses.
direGoldfish: I completely agree with you about curves. They are more difficult to select, they have no depth when viewing in 3D (a nurbs rig “reads” well even when nothing else is showing) and the color choices are awful. For something an animator has to stare at all day, it might as well look nice.
But since curves are easier to auto-script, more widely accepted and don’t cause any rendering issues, I’ve almost totally abandoned the nurbs technique. :sigh:
Chris: That’s all true. My auto-script still makes nurbs curves initially for the controls. The gui for my auto-rigger has three post-passes that each require an input file. One pass replaces all the curves with nurbs shapes that exist in another file, one imports skin data from a file onto the selected mesh, and one executes any character specific script I may have written for the character I’m working on.
Basically I have a library of control shapes from rigs that I’ve built in the past and I target them onto the new rig and tweak as needed. It’s fast-ish, but not automatic.
There are more elegant solutions if your auto-rig isn’t scripted. I know a number of people that have a library of re-posable rig pieces (limbs, hands, heads, spines, etc) that they stitch into a rig. There are a number of down-sides to going that route, but one of the upsides is you can easily connect the length of your limbs into the scales of your control shapes and never bother tweaking your shapes.
As for rendering issues, the rig gets blown away on export anyways, so that’s a less of an issue in games. The few times I’ve freelanced for small animation studios they haven’t complained about the nurbs thing… but I have no idea if the in-house TD changed things :X