Hi Bas,
If you take a 360ºx180º as result, then you have covered a whole view around you, no lens will change that.
The human eye has a total of around 180º, but a very narrow area of real attention. The perception of the narrow field is more for sharpness and color, but the wider it gets, the more it is more sensitive to movement. Typically we don’t notice how little this small “attention area” is, as we memorize things, and we have a very wide field of view for our context.
Besides that we can revolve the eyeball. To cover that, the viewer device/screen needs to have, and certainly not less field of view.
To understand the whole thing a little bit more, start with spherical VR.
You can use what ever lens you like (for a 360/180), on the end the point of view is the main part in it, and based on the lens (and sensor) you get a variation in the resolution, the parallax of the given 2D results of each “shot” will not change. You can’t make a 360º wider by choosing a wider lens, or narrower lens for that matter. If you lens is wider, and you stitch an 360ºx180º you need a lower amount of shots to cover the sphere. This is the base concept, if you limit the view to a fixed stereo (two sphere segments), the eyes can move while using a head-mounted-dispaly, so the angle of view from any given “eye-equivalent-lens” will “expand”. In other words, the field of view in the head mounting display is the one that is more important than the possible view with the eyes to create good experience.
(See also “Digital Stereoscopy” by Benoit Michel, page 49/50. He assumes a good fit for woman at 62mm, man at 65 and kids 48mm. Given his idea about, you have your answer about the distance.)
My personal take is, that the nodal point of each lens (eye) is always the critical factor. While we look at something very close (as in cross-eyed) the distance will change dramatically. I would avoid the gimmick effect of too close, it is weird and in comparison like an over saturated image, it doesn’t feel good on the long run.
The eye distance varies from human to human. There is not a “one size fits all”. What you notice it the effect of a scene size/scale disagreement with the eye distance. It can go in both ways, smaller or larger of course. So, you figured correctly.
That green boundary is a safety-zone. It doesn’t mean that it is comfortable for people starting directly after it. Yes, it should grow or shrink with a different eye distance.
All in all, you can ask a dozen people about their experience, how close things can come and how the scale was, or even how they feel (virtual motion sickness is a big theme), and you might get a dozen different replies.
Going deeper into Head Mounting Displays, there might be some distortion that could be counterbalanced, but the change might influence the viewing experience as well.
I would leave depth of fields effects and motion blur out of the equation, as that is not natural to the eye when one can interactively “look around”. A high fps number is key for quality, and a well though out presentation, and last but not least, the cut from one scene to the next is crucial as well.
The focal length by itself doesn’t define any field of view per se. The circle that a given lens projects can vary, comparing the same focal length. The field of view is not the key, the magnification is. (But yes, since we are sensor driven, we short cut this definition.) Lenses have no crop-factor (for example) sensors have in comparison to a given standard, typically 24mmx36mm. The eye’s magnification is obviously something that let us create the “sphere” as a representation of the world around us, a similar magnification is given with a certain lens, some say 50mm, some 55mm. What I took from Cognitive Psychology is, that we memorize a sphere like image while we rotate our head/eyes. Even if we bank (tilt) the head, this “sphere” stays leveled. So, many things to register to get a good VR.
All the best
P.S.: perhaps a good primer to most of the VR key themes is this pdf:
https://www.jauntvr.com/cdn/uploads/jaunt-vr-field-guide.pdf
