So… First of all: I know nothing about optics whatsoever, alas, but even somebody like each and every one of us here can intuit that if I have, say, a 180° FOV headset, and I can not reduce eye relief enough, when wearing it, so that there is actually a sliver of lens 90° out from my cornea, to look through, then we have a bit of a problem… :7
From what I have heard, the diopter adjustment in the recent headsets with pancake optics, which are all the rage at the moment, is done through moving the middle element in the three-lens/reflector stack.
(Also: I would presume the reported better stereo overlap in these headsets is down to greater magnification allowing for smaller screens, which means those screens will not bump up against one another in the middle, nearly as early.)
…but “all” we really need to do, for our single-lens devices, is to modulate the distance between lens and screen, just like when you focussed sunrays into a burn point using a magnifying glass, as a kid.
As I’ve mentioned a few times on this forum: The Oculus Rift DK1 came with three pairs of identical lenses, set into mounting tubes of three different heights, giving three fixed diopter options. Samsung’s ancient GearVR smartphone-drop-in HMD had a thumbwheel for dialling in the focus, so this is not exactly something new.
Even without an actual built-in mechanism, it should be perfectly possible to adjust lens-to-screen distance using shims under the swappable lenses, or e.g. screws around the perimeter, which act as “feet” for the lens, albeit only to increase the spacing – decreasing it, one would think, would require a more invasive procedure…
The only time you would need to have eye relief compensating in real time for diopter adjustments, would be if A) it is actually the lens that is moving/transforming, and not something else, and B) you are continually adjusting on the fly, to focus on whatever you are looking at, at any given moment, at the right focal distance, corresponding to the distance to the thing in the virtual world.
Oculus’ first “Half Dome” varifocal prototype used servos to move the screen back- and forwards. Later prototypes instead had the lenses suspended on thin leaf springs acting as parallel link arms, and they were actuated by speaker coils instead of motors, which was a much faster and quieter solution – apparently the variations in eye relief inherent to this was a relatively minor problem…
You would probably want the distortion profile to adapt to the modulating optical train, though – there is a bit of a “zooming” effect.
This also affects the eye tracking camera, which in the Reality series HMDs watches your eye through the lens.
It is possible the eye tracking calibration procedure could be enough to compensate for warping associated with changed diopters, as long as we stick with fixed focus optics. (EDIT2: …maybe even enough for entirely different lenses, such as the 42ppd option, or hypothetical third-party offerings, but that sounds less feasible.)
I am at a complete loss as to why wide FOV would necessitate a short focal distance - such a claim smells more than a little fishy to me…
So… No matter how well one may have engineered one’s lenses, using mathematical models; At some point the rubber must hit the road – that is to say, one need move on from the theoretical realm, and actually peer through the physical lens, produced in the real world, which one wrought by one’s calculations, and see what things really look like, and this is where subjectivity comes in.
…and this is where I could really see somebody with severe myopia experimenting, and determining: “Yep, this is it: At this distance between my beautiful lens and the screen, things look perfectly sharp to me – chisel down these numbers and built me a million units! -No, we don’t need a greater sample size, nor optical measurements using instruments.”
You might think this scenario preposterous, and that no engineer would be so cartoonishly lackadaisical, but I will remind you that this is the company that posted on this forum, several months after 8k/5k release, something to the effect of: “Hey guys. We’ve got some calibration equipment now.”
I’ve got my own little generic-lens-equipped “viewing box” (…or whatever one should call it), right here, which is entirely roughly “handheld-tuned” to where it “feels comfortable” to look into, for me, with my eyeballs. To anybody else, it would undoubtedly be terrible. But then - I am no engineer in any way, shape, or form, nor a businessman aiming to sell something to the masses. :7
About “cross-eye-ing” discomfort, I think we have discussed more than a few times on this forum, how the canting leads to the spot in each view that is best in focus, pointing 20° apart, so looking to the side brings one eye into focus, and the other out of focus. This wouldn’t be a problem if the whole view was in focus - then we wouldn’t notice, but it isn’t, and getting it all in focus is not an easy thing to achieve. (…although we do have some users who earnestly-seeming claim that is what they do see in the headsets…)
As for eye relief… Ye olde Rift DK1, again, had this through telescopic linkage between the HMD body, and the face gasket. So did the Vive, and so does the Valve Index.
…and I’ll reiterate my assessment, that for a wide FOV headset (…which I suppose leaves Crystal somewhat in the clear, in this case), one MUST have Z positioning.
EDIT: Heh, I wonder if one could get Jeroen of the Huygens Optics Youtube channel interested in VR, and have had the Pimax contingent drop by his place when they were in the Netherlands anyway, and leave behind a pile of lenses for him to evaluate…