Eyestrain discussion / Problems and Solutions

issue
events

#644

Risa’s explanation relies on the centre of the lenses not being the focal points. My explanation does.

For deliberation between both points of view you have to ask:
Why even have the smallest circle fresnel rings indicating the centre of the lenses if that’s not actually where the sweet spots are ??

Much more likely everyones looking at the inner edges of the sweet spots which fall within 65mm IPD since 70mm is the true minimum, and the diameter of the sweet spot is 2.5mm-5mm in each direction from the middle of the smallest fresnel rings.

There’s no assumptions being made, the data speaks for itself. Theres a discrepancy of 10mm extra measurement between the physical lens distances and the on screen IPD setting, on screen IPD being 10mm less than physical distance.
Mathematics and geometry are saying it should be a smaller lens distance created due to divergence of panels and lenses, compared to a 180 degree plane and that should be accounted for with extra measurement on the software IPD reading to give an accurate reading, not a reduced software IPD reading.

The middle point of a line on a 45 degree angle is shorter than the middle point of a line on a 180 degree angle, which is why extra software IPD measurement should be added to compensate for the divergent displays to give an accurate 180 flat planed vr IPD reading (which corresponds to real IPD), not by reducing the on-screen IPD reading. This is my explanation for the IPD measurement discrepancy being 10mm less on screen than the 70mm physical minimum distance between the lenses on Pimax.

I believe the measurement discrepancy is caused by Pimax opting to reduce IPD to increase lens real estate as much as possible, and why the minimum distance between the centre of each lenses is 70mm, whereas most other 100 fov consumer vr headsets have a 60mm minimum IPD and IPD values which correspond to the users IPD measurements.

I’ll draw a diagram and post a thread later if you still don’t understand.


#645

There is that think word. Well soft ipd already works on fixed lenses. We know pimax uses both. So an off set should be tried before thinking it won’t work. Especially since it has been proven to work in other applications. Knowing what part of the lens each eye is looking through is key.

We won’t know well or not it works until we do.

The rings will alter the distortions for sure. Fresnel is a compromise and hopefully something better gets designed. Eye projection though I see being the future with tiny lenses.

That or the neural implant


#646

Future design should have no knob. Embedded Eyetracking with auto ipd self adjusting leneses. :beers::sunglasses::+1::sparkles:


#647

Do you think that would impact durability? Compared to an analogue knob an automatic mechanical adjustion calculated by software sounds like the varifocal display tech oculus had to stop because of durability issues with the extra moving parts.


#648

Varifocal is constant moving(so to speak). Self ipd would only be needed for initial setup/different user.


#649

Exactly what is happening, In risa case if he looks left by 3-4 degrees he now has his focus in his left eye because he is now looking though the center ring but his right eye will move more out of focus and will be looking thought about the 4 ring. that is sown in his diagram. I agree that eye tracking can fix the stereo 3D image but it cant and will not fix the out of focus lenses.


#650

Risa’s explanation does have the centre of the lenses, extended, being their focal points, passing straight through the pupil of the eye. -It is just that since the lenses, and the screens with them, are rotated, those focal points are 10° out to either side. The (say 70cc) lenses point in, straight at the (say 60cc) eyes; Hence the “discrepancy”; The displayed IPD value is for the distance between the user’s pupils, not between the lens centres. The diagram illustrates this perfectly.

Meanwhile, your explanation, as far as I am getting it, leaves no direction you can look, where your eyes are aligned with the lenses. You do intersect their centres, which gives you as decent-ish simultaneous focus for both eyes simultaneously as you can get (given that you are still looking diagonally through the lenses), whilst looking straight ahead, but throw away a litter of babies with the bathwater.

You do not get better sharpness at the centre of the lens becase that part is inherently some sort of “sweet spot” in itself, but because that is where the tangent of the focal field curvature of the lens touches the flat display panel – the farther out from the centre we get, the closer to the lens the screen would need to be (EDIT: …relative to where it actually is), in order to remain in focus – a problem for which several potential solutions have been proposed (and patents both approved and expired) over the years.

??? The purpose of that ring, nor the others, is not to indicate anything - they are just the segments of an equivalent conventional lens, that have been collapsed into a thinner cross section.

Again: When you look 10° to the right, your right eye will be right in the sweet spot of the right lens, receiving optimal focus and distortion mitigation (provided your eye-to-lens distance has been adjusted appropriately), and when you look 10° to the left, the same will apply for the left eye.
Consequently, there is no place where you can get a perfect perpendicular view through both lenses at the same time.
This is the tradeoff chosen for the Pimax 8k/5k, given its FOV design target, and its following screen cantering, whilst retaining relatively simple optics, for better and for worse – and I have certainly spilled my fair share about the “worse”. (something quite similar was, as it happens, elected for use with the Oculus Rift CV1).

Just to reiterate: The diagram of Risa’s, where he has both eyes looking forwards and both gazes intersecting the centre of the lenses, is a demonstration of the wrong way to fit the HMD. -If one do it, one will indeed get a little bit better focus focus, in stereo, when looking straight ahead, but both lens distortion, and the software distortion applied to counter it, will be skewed, and the focus (often misattributed: “sweet spot”) you gain when looking straight ahead, will be robbed from when looking even slightly to the side.

Just so that we are on the same semantic wavelength here: Your definition of “sweet spot”? It should be the place in front of the lens, where your pupil needs to be, for optimal optical conditions, but in VR circles it has been widely misappropriated for how far one can look, before field curvature makes things too blurry to read. I’m kind of getting the impression you are trying to coin a third definition of a certain area of the lens, which would be unfortunate.

I’m going to have to request graphical and mathematical presentation of all that reasoning, I’m afraid. “Middle point of a line on an angle…” – context please.

Well, yes-ish… (If I’m following you correctly) – there is no discrepancy, as explained above, but there is more screen that needs to be covered by the lens, and you want to assign some of the lens to it, other than the least tractable bits in the far periphery. :7

Since the axes of the lenses in HMDs like the first Vives and Rifts are parallel, the Inter-lens distance follows one’s IPD directly, but with the canted lenses, they converge on one’s eyes, on the outside of the HMD (…which is the same as the lenses diverging, from the point of view of the eyes), and the relationship becomes a function involving the amount of eye relief.

Please do.


#651

But it will to a point. If his one eye is not looking through the center either his eyes have a wider Asymetry or he has shifted the headset so one eye is looking through the center. If the headset is rendering via symetry then it is basing evem distance 32mm×32mm for example. Then for Asymetry it would need offset(s) to render a proper image based on actual eye position. Say 31mm×33mm. L @ ring 2 & R @ ring 4 Rings eyes both on the nose side of sweet spot.

Eyes get clearer view looking towards ears of coresponding eye.

Like @jojon said & @Cdaked posted long ago was one solution is a rendered nose.


#652

Posted a diagram to scale in a new thread as to show you how divergent lenses reduce the minimum distance between lenses and minimum IPD reading, which is the primary basis for my reasoning as to why there is in fact a discrepancy when the on screen IPD reading should be increasing, not decreasing, to compensate for the distance between the lenses lost using divergent or angled screens compared to flat-plane screens (the IPD measurements of which correctly translate to measured far IPD, such as on vive or oculus.)

Thanks for your detailed response though. Identifying the real cause of the IPD discrepancy issue: Does Pimax intend to offer a solution besides a software fix?


#653

And yet, with my IPD of 63 mm, I can play for 4+ hours with no eyestrain at all. Respectfully, I think there is something more going on, than just an IPD issue. My understanding is that since VR headsets aren’t displaying a true “light field”, that vergence issues will cause eyestrain for some people on just about any VR headset. (Note that I am NOT saying this will be the same group for all headsets, some may have problems on particular headsets, while others may have problems with other headsets.)

There’s no doubt that Pimax has a unique design and that some people are having issues with it. My point is that this might not (entirely or primarily) be due to the IPD and lens design; other factors may be at work.

If, as you say, I should reduce my IPD setting, from 63 mm downwards and look through the centers of the lenses. Yes, I can do that and the forward view will be slightly more clear. However, that causes some distortions in the binocular overlap region and quickly causes eyestrain. My take is that there is some sort of head-shape/eye-position limitation (per individual) that might be due to something other than IPD settings.


#654

So what are your saying exactly? That you want things to line up the way they do in the lower half of your first picture?

If so: That is exactly the mistaken idea (“near IPD” and all that), that is the cause of your discomfort.

Some focus and geometric accuracy when looking ahead, has been sacrificed for some focus and accuracy to the extended sides. This is itself ís a bit uncomfortable, but not nearly as much as the alternative. -Roll with the punches; Don’t try to strike them back with your chin.

I should probably mention that I was asking (unanswered) about this very matter, and quite a few others which were also predicted, during the months before shipping finally began. The current situation was one on my medium-worse-ish case scenarios, but I can appreciate the situation, grok the technical issues on a layman level, and adapt.

( EDIT: For reference, my IPD is somewhere around 59-60mm, and I am not only quite comfortable in my 5k+, but given the fact that I have rather protruding cheekbones and forehead; If using the foam mask that was delivered with my 5k, I actually have to increase the IPD dial a fair bit higher than my 59-60-ish IRL IPD, to compensate for the high eye relief – any lower, and my eyes begin to cross. )


#655

Near IPD measurements are -3/-4mm your far IPD measurements. Not 10mm+ your far IPD measurement.

I’m still boggled and curious as to why Pimax would not intend the centre of their lenses be used as the focal points for both eyes, the fresnel ring placements tell me a different story as to the intended focal point of the lenses.

I totally understand how not looking through the centre of each lense compensates for the 70mm physical minimal > 60mm on screen readings; however, that understanding relies on the centre of the lenses not being the intended focal points for each eye.

@Dallas.Hao @Sean.Huang @Doman.Chen @Alan.sun @PimaxUSA @mozi would Pimax care to elaborate on what the intended focal points are for the lenses? is it the centre of the lenses or inwards towards the nose from the centre of the lenses by a few mm as @risa2000 suggested?


#656

As always: Tradeoff. You give some to get some.


#657

Why not centre the fresnel rings around the actual focal points of the lenses if this was the case? A few milimetres closer together than the fresnel rings actually are? Those rings are designed to indicate where to look at.


#658

Reread my long bit o’ waffling, above.


#659

If my hypothesis explaining the IPD measurement distortion is correct, and people are already looking at the outer edges of the sweet spot and barely scraping by with clarity in both eyes, wouldnt centreing the lenses create more stereo overlap when looking left to right by 10 degrees simply because you’re staring at the middle of the sweet spots so you have both directions of diameter to diverge your eyes from?

As it currently stands if you look left or right there is no stereo overlap, and I believe one eye definitely goes out of the sweet spot when looking left to right with the current design; whereas, with slightly smaller lenses allowing for true 60mm minimum physical distances, that may place peoples eyes more in the centre and allow for stereoscopic vision left and right within the diameter of the sweet spot for each eye (not sweet spot binocular overlap)


#660

I think I actually get less eyestrain with my 5K+ than I do with my Vive Pro.


#661

Really simple if 70mm is the space for 60 ipd than if 10mm adjustment is direct than at 70ipd the ring distance should be 80mm. Either way 70ipd will have a greater space than 60ipd. Meaning in theory no ipd is intended to look through center of sweetspot. But just off to the side closest to nose per eye.


#662

Question of design to desired effect with trade offs.


#663

A or B, I am using A, I can not use B.