I have thought about this question plenty of times and have not come up with a decent theory.
The weapon hit detection is to my best knowledge tied to the actual placement, shape and orientation of the weapons, but even if so, it still doesn’t make sense to me.
There are always complaints over enemy attack range / visual disconnect and also own weapon animations not being accurate for aiming at heads etc.
Then the newer issues of visibility behind shield weapons and a bit rarer discussions on how ranged weapon zoom (especially with WS) look really odd.
I feel all of these have at least part of their roots in the 1st person arm / weapon models being at wrong scale (unless you play at min FOV which looks like it matches then.)
If the own model was rendered at the same FOV and scale as the rest of the scene it would automatically make enemy attacks look as if they actually connect (due to the model ”shrinking” toward the center of FOV). Also own hits looking accurate could actually happen if the colliders were as closely as possible following the own model extents and joints. No longer would the models be bigger than Lohner for example and would clip into geometry nicely and not look off / wrong distance and scale.
Visibility over shields would improve as you gain real screen space around the models unlike now where by increasing FOV it looks as if you are walking backwards with something weird happening with perspective.
Should / could the FOV of the 1st person models follow the main FOV (and zoom even)?
If not, I’m interested in hearing why it is how it is as I can’t figure it out.