I'm just now starting to tackle learning rigging.
When I do an internet search for T pose or A Pose, I see the legs are either posed side by side as you would naturally stand, or further spread apart as if the legs themselves were an A pose.
I would think the more natural pose would be better, but if that were the case, why are legs sometimes spread further apart?
It also seems the legs being spread further apart is correlated with the A pose for the arms. Does how the arms are posed affect the legs, and if so, what are the mechanics that are going on?
Thank you!
Replies
In the simplest case , you want the character posed in the middle of it's most common range of movement to minimise the amount of mesh distortion that occurs
If your character spends 90% of the time walking round with a gun in their hands you don't really need the full Tpose and you might as well keep the legs fairly close together.
If the character backflips all over the place you probably want a more extreme pose.
I like to think of the poses as a little more gradual and fluid, not locked into "one pose to rule them all". The whole point of being able to skin up a character is so that you can pose them, so absolutely leverage that and help out everyone at each stage of the pipeline.
Rig pose : The pose that you use to create the control rig doesn't even really involve the mesh, its just the skeleton and controllers. You create joints and hook up control shapes, create the motion systems that drive everything (IK/FK, driven keys, constraints ect..) this doesn't necessarily need to match your mesh perfectly. Rigging is easier when things are mostly straight with predictable default values. So using a T pose with feet facing perfect forward and arms facing away from the body at 90 degrees, straight fingers, it is great, for rigging.
Skin/Bind Pose : This can be whatever shape that matches your actual character. Baked in a motor bike pose? Great, I don't really care, I'll skin in whatever pose we need but I'm not creating a rig in that pose, unless I'm using a tool that helps with that.
Another example is bake vs bind pose, especially when it comes to faces. It helps to bake faces with the mouth slightly open and the eyes slightly closed, but no one during the sculpting approval phase likes looking at stoned characters, so you pose it, bake it and put it back.
Animation Pose : This is a default pose that animators like, usually this matches the bind pose, but it doesn't need to. Maybe the bind is a A Pose but the mocap system likes starting from a T pose, so give em what they need to succeed.
There are advantages and disadvantages of changing the poses or keeping them similar but it all depends on the tools you're using and the pipeline you've got going. There are ways to deal with every combination of tools and poses and none of it should completely brick your pipeline even if you pick the worst combinations possible. So you're pretty much free to do whatever works best for your team and your project. If that's everyone agreeing on one pose and everyone working around it, great. But I don't really see that as a necessity.
Personally, I think you should vet your pipeline with block out assets, make sure that all works exactly like you need it to, and then go about putting in all of the work to make the final asset. It helps to do it in stages and refresh the meshes as you go if someone has questions about a specific asset, like a crossbody shoulder bag or whatever, you can test it out before you dig in deep on asset creation. Take your time to figure out your topology and get things working like you want before artists spend all of that time.