Something I've often wondered about is the "bind pose". The bind pose for human skeletons is one where their legs are straight, and their arms are perpendicular to their body:
My question is: why? Is such a pose easier for the animators to work with, or the programmers, or both? As far as I understand the math, the bind pose doesn't make a difference to the outcome.
It's easier to model because there's separation between the limbs and body, and the bits that need to stretch when animated are already stretched - so they get modelled and weighted correctly.
Same reason you'd model a hand as open and fingers spread, rather than modelling it as a fist.
Oh, and as far as programmers are concerned, we don't care. It'll still look like something from "The Thing" when we get the skinning code wrong at the first attempt.
As for the reasoning behind the pose. There are three.
As to the T-pose vs. 45-degree inverted Y pose.
When you use skeletal deformation for animation you are using bone positions to create deformations in the shape of the model. This approximates the motion of limbs, but it's only an approximation, and the further from the bind pose you get the more distorted the deformation. There are lots of expensive solutions to fix that, creating target shapes that you blend the verts towards as you use the bones to move the model but those solutions get expensive in memory or cpu time quickly.
So why the "T"? Well if you look at the range of motion a human arm can make it goes from a little past pointing straight up to a little past straight down. This makes out to the side exactly in the middle of the two extreme positions. Since skeletal deformations look worse the further they get from the original pose it makes sense to make the bind pose in the middle.
As FuzzySpoon mentioned the static-T pose has been shifted a bit to arms at more of a 45-degree downward slope. The reason for this is that while horizontal is the middle point for the full range of rotation of the arm, it isn't the middle point for common usage of the arm. Our arms spend most of the time either at our sides or as high as shoulder height. So 45-degrees down is right in the middle of the most common positions for arms in animations. The other issue is that as you lift your arm to horizontal your shoulders tend to rise and clothing bunches up, it's more complicated to model the shoulder area when you have to think of pre-bunched up clothing and raised shoulders.
My view is that things are changing often. People get better, and people fall behind. The best place to look for reference are the people who are kicking ass at whatever you are interested in. Lately, the traditional static TPose has been less favoured due to IK/FK and other animation problems when rigging in such a rigid position.
Here is a great reference : http://www.zbrushcentral.com/showthread.php?page=1&pp=15&t=79141
Notice their TPose character shots are all "loose" and far less static. The elbows are bent slightly inward toward the body and it allows a lot more natural flow on the modelling of clothes, and skinning becomes a lot easier for an artist. The legs are a little bent as well, and the shoulders are not high up there they are more natural for when the animation comes in. It allows a much better topology flow to be created without losing the simplicity of rigging in such a pose.
And to answer the question directly : Its for animation and rigging purposes. It's a horrible workflow trying to rig bones in the spine if the character is sitting down. Or even in the knees, if bent. When painting weights for things like materials and cloths you will find this sort of pose to enable a much quicker workflow.
