Dumb game graphic question.

[spoonsinger]

Why is T-Pose a thing in so many games?  I understand its a fault condition, but I would assume that many different games use different development environments and would therefore have a different default fault position. Or is it that all games which show this condition use the same base tools, (or variants thereof)? Or is it just a convention for skeletons as a whole. Just curious.

[Just Don't]

https://en.wikipedia.org/wiki/T-pose

 

It's the default stance in most tools to start animating.

Also a common placeholder and even a required step if using mocap.

[Papitas]

I wish I had sources for this, but I read about it many years ago. 

 

Most humanoids are modelled/sculpted/whatever at T-pose (and thus, making it the default position when no animation is found) because it's easier to make the topology for the armpits at that pose. If you ever try to make a 3D model, you will notice it's a pain in the ass modeling the pubic region; the T-pose avoids that problem at armpit level, which is way more important to get right, since shoulders have way more planes of mobility compared to hips-legs. 

 

I've seen artists arguing about T-pose is not actually the best pose for modelling. They say having the arms lowered (at 45° instead the 90° from the T-pose) translates to better deformations. But I don't know about that, though. I haven't tried that myself.  

[solo.kirmish]

If the crotch is such a problem, why not use the Da Vinci pose instead?

 

But the real issue is with all body parts who have skin stretching and skin rotating/revolving going on, because you have difficulties foreseeing how they will look after this extensive manipulation is done. Our interpolation techniques for stretching deformations used to be really bad for a long time, and are still not perfect, although we have made some substantial progress. If you are interested, you can look up Quaternion Rotation. Here out of an old paper I had:
 

Spoiler

 

 

 

This is why it is often not so intuitive to model the regions in questions, because though they do need more details to make stretching realistic, if you do too much, it will also negatively affect the rotation calculation. It is difficult to figure out the best solution ad-hoc. Trial and revision is still often the only way I believe. (maybe modern industry has behind doors advanced beyond this by now, not sure)

[Nonsequitur]

Ever "start from Scratch" after spending months on a project? 

Imagine having the entire game and supporting industries throwing everything out and starting over... 

 

Too entrenched.

 

[27X]

22 hours ago, Papitas said:

I've seen artists arguing about T-pose is not actually the best pose for modelling. They say having the arms lowered (at 45° instead the 90° from the T-pose) translates to better deformations. But I don't know about that, though. I haven't tried that myself.  

Correct.

[Nonsequitur]

t-Pose is better for seam alignment of object components dealing with material physics. 

However, there is a trade-off. With lower resolution frames/materials, the bending tends to clip & deform more with the more radius or bending you have (depending on number of faces).  Sims 3 and many other similar systems are very low Poly. So, went with the modified T-pose as it is closer to the natural state - less overall deviation, less face movement.

My thoughts...