even after further testing i can’t get to a point where it feels comparable to a Vive
the problem is not what happens in the centre or outer area its about around the nose (overlap) and the lower area and i think i found what feels uncomfortable
when coming near to the inner edge of the lens things keep there vertical geometry (at least its possible to find a position where it is the case) but things get compressed, get thinner and that makes a difference between left an right eye, the other eye perceived the object in its normal "thickness" so if you move the head left/right the moment when perceiving a object changes from one eye to the other they have different geometry and it feels wrong/unnatural
kind of same happens in the lower area when moving the head up/down
i took steam vr home as reference, standing in the room and looking outside, the wooden frames left and right and the tree can be used as reference (also the 3 pictures on the wall can be used), close one eye, move head to look near the edge of the lens, move left right to see the thickness of the wooden frame and tree change, now use the other eye, same position same "object" but no difference in thickness when moving slightly left and right
I’ve done the same with my Vive (gear vr mod) and there is no such extreme change in geometry on the edge or near the edge, it just gets blurry on the edge but the objects shape stays intact, so when perception changes from one eye to the other there is just a change in sharpness, resulting in jus a little blur in that area when moving
with the wooden floor panels it can easily be seen when the geometry changes when moving the head up/down, not a problem with the Vive
would be interesting so see if other people see that too with there pimax and could compare to a unmoded vive
I haven’t noticed this. Perhaps facepad thickness (eye distance from lens) is a factor. The game itself is also a factor. I see some distortion on the outer edge in the SteamVR demo room (with the wood paneling and the outdoor view). I see almost no distortion in the Elite Dangerous lobby (a spaceship hanger). They are both rooms, so I’m not sure why it differs.
can you reproduce the compression/warping in the steam vr room?
just a closer look to the edge and you know what i’m talking about
its not the room/enviroment as the same can’t be seen with te vive
using different enviroments is not so good, but i guess something not to dark with objects that have a clear geometry should show it too (small leaves of a tree moving in a breeze are not qualified as test objects i guess)
also interesting if someone with q WMR hmd could test (OD/OD+?)
when i’m right you can use any enviroment and software as long as it has objects with a clear shape big enough
should also be independet from fov setting, use small if you want the distortion correction in the overlap aread around the nose will be the same
Have you tried spacing the lenses a little farther than your IPD, to account for the canted screen and lens assembly, so that the lens axes intersect their respective eyeball? (as illustrated in another thread, by Risa2000)
Several downsides to this, and nothing is going to fix everything, for evey situation, but… There are a whole host of things I feel inclined to add, expand, and further speculate upon, but too much is too much.
no, problems in the outer area a pretty normal and can be minimized by tilting the hmd (for most) but the problem i’m talking about had no impovement for me whatever i did
prety sure some (most?) people can adapt (the 3 testers never mentioned that, but i can remeber a few people in the berlin meet up videos mentioning slight weirdness in the nose area)
at least it was something i could nail down why it feels off for me, I could compare it to the Vive and see the difference - much larger edge area with distorsion with pimax
yes, my ipd is 67 and a tryed it up and down
(also removed facepad and tryed every eye-lens distance from as close as possible without foam to where you can see both lenses with one eye, also tilting upper and lower part - does not change what happens on the inner edge of the lens, can only make it worth by adding vertical distortions with straight lines bending left/right
please try the test i wrote about, i’m interested to know if anyone who already find his optimal spot can see it too (and it does not bother him) or if he does not see that compression, also most backer will have vive/vive pro and can cross test themself (as is did)
I’m about to go to bed, but I’ll see if I can find some time after work tomorrow.
However: I very much recognise the distortion toward the edge (EDIT: in this case highlighting the “inner” edge) of the lens, that you describe, and the binocular rivalry it causes due to the compression of the view there (along with blurring and ghosting in the fresnel lens segments) in the opposite eye - in our case exacerbated by the particular canted screen-and-lens construction of the PM8k/5k.
Perhaps we will get an updated lens compensation algorithm, that helps a little with the problem, by utilising eyetracking to compensate appropriately for pupil swim (currently things in the periphery look pretty correct, if you look right at them, but if one looks straight ahead, they compress), but for now, I believe the best we can do (barring reconstructing the HMD :P) is to: A) Find the sweet spots of both lenses, taking into account that these will be with you looking toward each lens - 10° out the the side, rather than straight ahead (something the Vive and Rift are not subject to, having their projections parallel), so that we get the best out of the static lens compensation, even if we’ll never get both eyes perfectly in focus simultaneously, and B) Get accustomed to the warping, like anybody who gets a new pair of eyeglasses.
Yes. I just retested the Steam VR waiting room. At my measured IPD (63 mm), my normal VR IPD setting (65 mm) and the maximum IPD setting, I see no inner distortion. Pulling the headset forward (to simulate thicker foam) made no obvious difference (to the inside). When setting the IPD to the minimum value, I did see some very slight inner distortion. Also of note, at the max IPD, I quickly started to get a headache.
Keep in mind that I wear progressive multifocal lenses (in real life, not in VR), so I may be less sensitive to distortion than others.
Heh, kind of relieved in a way, that I’m not the only one who’s ending up performing little mostly faith-based rituals, to placate the work-dammit gods. :7
so you see the same problem as i do, but choose to ignore it (as most here i guess)
why? i don’t see a difference to a normal gen1 hmd (like vive), you look with one eye trough a lens, lens and panel are parallel
if pimax would have done it right it should look like in the vive, the main difference for the pimax’s higher fov is the outer area, the inner area is just the same tech as with the old gen1 hmd’s
the question is, sloppy measuring (lens profile, can be corrected) or a problem with engineering
if they did not correct it in the static lens profile, why should they bother in a dynamic, its just ignored or not properly analysed
the inner edge problem is not so much about pupil swim, its not ok when you look straight and its not ok ok if you look direkt to the edge
we dont know how pimax creates the lens profile, they will have to choose one distance lens/eye (lens/camera) aka eye relief and then they dial in a ipd on the hmd and take that value as distance between the two cameras?
its not a problem like the outer area where there is a huge difference between looking staight and looking into the outer area, this problem does not exist in the vive so it can be solved
making a 0.75 step foeward for a wider field of view and making 1 step backwards in the “normal” area of vision is not the way to go
i’ve tryed very hard but i cant get rid of the objects getting thinner in the inner edge area, at the spot where at least the vertical geometry stays intact there is warping in the lower lens area, when correcting this by tilting the hmd’s upper half away from the head it add’s vartical distorsion to the inner area
as i’m the only one bothered by that it looks like i should chew some ginger and get used to it
i guess i will have to keep my vive in case i want a “relaxed” vr session or have someone else over trying vr
at least i can still hope valve has something to offer later this year and i guess they wil have more sense in making the experiance enjoyable and not fighting your body (using the vive felt not like hard work)
i guess you would have to use very thick glases to see anything like this, i use glases for reading in short distance, there is no warped image between my eyes
I see it, and it annoys me greatly, along with many other issues (couldn’t comment on behalf of those who claim to notice nothing whatsoever, but we’ve repeatedly been dealing with those, all along (“no SDE in Rift DK2”, anyone?)), but I never really expected better (go back, and you’ll find plenty of my trying (fruitlessly) to temper people’s expectations, and pointing out obvious technical challenges), and I can roughly understand technical limitations.
…so yes, I do choose to, to a degree, overlook something I both foresaw, and do not expect to see vastly improve.
Ok, so posit you are looking at something that is a teoretical infinite distance away (so that your left and right sightlines are parallel - no convergence), right in front of you, wearing your Vive: Here you are looking through the centre of both lenses, along their axes.
Now: Given that these lenses are in-line, perpendicular to your straight-ahead sightline, and perfectly symmetrical: If whatever you are staring at now moves a little to the side, both eyes, tracking it, will look at the same spot on their respective lens, resulting in similar optical situations for them.
This is not the case with the Pimax 8k/5k. -The screens, and the lenses with them, have been swivelled out ten degrees to each side, but the lenses do not appear to have any prismatic properties to account for this, and still look pretty nice and symmetrically round, so the axes will have moved right along with them.
One of the consequences of this, is that when you look straight ahead, as in the previous example, you are not longer looking along the axes of the lenses; They stand slightly “on end”, so to speak, so you are instead looking a bit diagonally through them, which is not optimal.
Given that one lens/screen is canted to the left, and the other to the right, it also means that things off the perfect vertical centre line of will be different between the eyes; Something that is a bit to the right in the virtual world, that would be seen through roughly the same spot on each lens in the vive, will be right there, and nice on the right eye lens on the 8k/5k, which is canted to the right, but be about to “slip off the edge” of the one to the left, that is canted leftwards.
(Really: Look at Risa’s posts on the subject. He has nice sketches, that graphically represent the situation)
For a static lens compensation distortion, Pimax then, like anybody, has to choose something that works decently most-of-the-time, despite the user wanting to look around, within the headset. They will, from what I can tell, have stayed with the same sort of model as they used in their Rift DK1-derivative Pimax4k, and you have with the Vive and Rift; A barrel distortion around the axis of the lens, without any adjustment for off-axis viewing. This sacrifices some clarity when looking straight ahead, for increasing the apparent area where things look pretty much legible (EDIT: …a given bit being sharp in one eye, at least): Oculus has made a similar tradeoff with their CV1.
Only way to look (EDIT3: …simultaneously…) perfectly along the axes of both lenses in the 8k/5k, for optimal clarity and distortion in both eyes, is if you are walleyed, with each eyeball toed out 10 degrees.
Now, neither the Vive nor Rift is free of similar problems - not by a longshot (and I find plenty of different kinds of discomfort in both), but given the circumstances described above, they suffer less. Now, they too could benefit from…
…dynamically adjusting the barrel distortion, to counter pupil swim, when you are not looking through the lenses perfectly through their axes. This is a much lesser problem with the Vive and Rift CV1, partly thanks to the parallel projections mentioned above, but mainly to the thinness of the lenses (this is why fresnels were chosen for gen1 over conventional lenses, despite their many obvious downsides), but it is by no means eliminated.
For the 8k/5k, it could (assuming the eyetracking is good enough to begin with, and other factors not limiting), definitely help deal with the distortion changing depending on which way you are looking.
Finding optimal positioning remains a problem (and not just with Pimax, for that matter), especially with slippage, but it is not impossible. There are also things like an eye relief parameter, to the barrel distortion algorithm, which are not exposed to the user, but I guess we’re left having to try to add/remove padding as needed (EDIT4: …to match the hardcoded one).
Overall, I find my 5k+ overall a not insignificant improvement over both my OG Vive and Rift CV1. I miss the amount of stereo overlap, brightness, contrast, and black levels of the Vive (EDIT2: not so much the Rift), but other matters outweigh that loss to me, despite plenty of new shortcomings.
I too, am extremely keen on hearing more about the Vive Valve HMD, assuming it turns out to, at all, be something they intend to make a product out of, and what pros and cons it may have. :7
First time I put on a pair of +1.5 reading glasses, the entire world swam before my eyes, and I had to sit down for a while and breathe. I’ve gotten used to using them now, but absolutely still notice obvious warping, even in the ones I’ve found with the least amount of it – I see it with sunglasses too, for that matter.
I get this, for me wedging a finger under the headset over my cheek (tilting it up and away from my eyes slightly) makes all the distortion go away. At some point I need to mod the cushion to give me that all the time.
