3D Setup
its not about 'better/different' than what our eyes do. Its about maximizing the depth available in your field of view.
Imagine you get the best depth experience in the area 1 to 10 ft in front of your eyes. Beyond that region, you loose depth perception, and things only look 'far' away. But no longer really have a true sense of depth (its easier to judge is something is 1 to 10 ft away, than judging if something is 500 to 510 ft away)
By varying the distance between the lenses, you can effectivly 'scale' the scene (the virtual 'head' the cameras are attached to, grows). Thus making the 1 to 10ft in front of the eyes thats 'in focus' and showing maximum depth perception, suddenly go to 6ft to 100 ft of depth. So things to close to the camera dont work anymore, but for bigger scenery, or imagery further away, it creates the depth across the whole scene.
Combined with the camera spacing, the lenses also focus on the subject. Similar to the eyes.
Something to close (outside the focal range) will make you look cross-eyed.
Take a look at
this video, for the more expensive setups. But the same concept.
The ones we normaly use have a dual setup. One for close up shooting (1 to 10 ft) and one thats wider appart (30 cm or so) for stuff further away.
Its possible to shoot with one setup, but its just a matter of experimenting with the range that one gives you. Idealy your setup allows for two (near/far) setups, so you can do a 'close up' shoot on one jump. And a 'far away' shoot on the next one. So you have a nice range for editing.
The cameras should always converge towards the same center-point. So for the tandem-example you showed, that one would probably be 3 ft away or so.
There is also a lot of post processing you can do with your footage, to alter the focal point. But thats another 10 posts or so...
When working fully in the virtual realm its a lot easier (though using the same principles).
And there are also some ways of creating 3D in post, deriving it from left to righ/right to left motion in source footage.
An example of that one can be seen
here (the wingsuit footage in the 2nd half was shot with only one camera, and post processed to create a faux 3D effect, the rest is 3D animation on photos etc.)