PSA: Reverb G2 small sweet spots, observations and solutions

Yes these are Foster Grant.

Alden Reading Glasses for Men | Foster Grant

R: 52 [] 19-140 PD58 5MM +1.00
L: MS0320 Alden BLK

NB: the [] is printed as a :white_square_button:

update: they do not offer +1.00 correction on the web site!?

update: reading the specs online this means the indication on the right is meaning:

52: Lens Width
19: Bridge Size
140: Temple Length

NB: the 3 values above are matching the ruler on my photo!

PD58: I suppose the glasses IPD is 58, I’m 64, but this shall be good when verging for reading and therefore begs the question when used into my headset.

5MM: I’m not sure but could be lens thickness

1 Like

Hold on with this, I’m trying another pair of glasses which are incredibly better, I’ll post more about this in a few.

Update:

I’ve now tried another model which is much more comfortable on the bridge for me (no pads pressing the nose):
Paolo Semi Rimless Reading Glasses for Men | Foster Grant

R: 52 [] 17-140 PD62MM +2.00
L: CT1110 Paolo BLK

I’ve also measured the FOV and all the other tests with:
Steam Workshop::ROV Test FOV & Resolution

I reach HFOV 98 (didn’t bother with VFOV) and all the panels are clear of any CA and other deformations. I can nearly read left to right on a 85 deg from center (albeit fuzzy a little bit toward the edges as expected but legible enough yet). I believe these have a lens IPD of 62 and since I’m 64 this might explain why they are better. (I never paid attention to these markings, thank you for asking!!)

These glasses are better then he previous ones in comfort and clarity to me and this again tells me the G2 optics are designed for wearing glasses for glass wearers.

When using the Frankenmask you’re only putting your eyes in a different focal point which is compensating but at the expense of creating more distortions. These distortions might not appear as pincushion or barrel, but as white lines dedoubling because of a 2nd factor playing here: the WMR anti-CA filter which is pre-distorting the RGB planes so that lenses refraction put them back together.

Please note if I rotate the headset something about maybe 2 or 3deg pressing on the nose and pulling out the forehead at the same time, and if I adjust the headset to put my eyes slightly above the G2 lens center (you can see where on the fresnel) and without wearing glasses, I can read a clear view without much eye effort but with yellow and blue fringes showing up around dark lines over white. I can reverse the angle and then put the eyes on the opposite offset vertically from the lenses centers and again I can see the same clear with fringes. If I put the eyes back over the fresnel centers, no matter the angle nose pressing or forehead pressing, the image is less sharp but there are no fringes.

1 Like

On batches vs. HP Reverb G2 quality, Sebastian of MRTV has a nice, limited-in-scope review comparing the quality of a review model received direct from HP (you’d think they’d be sure to send the very best unit) with a G2 randomly bought from Amazon-Spain. Sebastian uses the Steam Workshop ROV Test utilities you recommend, too. He concludes that both G2 have superb optical quality within the limits of VR headset clarity and discusses issues that can affect the size of the sweet spot, including head size and the possible need for the Frankenmod. Commenters on his YouTube post suggest what he really needs to do is to test G2 units from people who are complaining about sweet spot size and quality. Sebastian also says be sure to be running the latest Windows Updates in there has been a problem with chromatic aberration for WMR headsets and in his humble opinion the latest Window Update takes care of that for the G2, at least. Sorry if you’ve already covered this in other threads. I have a lot of reading to do! Sebastian also has a lot of good stuff on his site, as well.

Maybe I need to install SteamVR, etc. I haven’t been able to figure out how to download and run the ROV Test FOV & Resolution utility yet. I do have Steam and Steam FSX on my computer running fine.

Thank you for the link and info! I know his videos and as a matter of fact he his also active on the reddit HP reverb sub and there are some good comments there (but not as friendly as here!)

Once thing remains for sure: I get terrible eye strain with the G2 because I have to force as when I’m trying to read something closer, whereas I can play Alyx with the Index for hours and don’t feel any fatigue at all… Optics are really different between the two and the more I think about it the more it makes sense. For example see this comment from the HP guy behind the G2 tech:

kaiserkannon HP Employee 2 days ago

I love high fields of view! But why did we choose not to expand FOV is a great question.

At a basic level there are two ways to increase FOV assuming same technology is being used:

  1. increase panel and optics size (Index uses big lenses on big panels). This comes with the con of increased weight, size, and cost.
  2. increase optical power while keeping displays and optics the same size. this comes with the con of lowering optical performance.

At the end of the day we did not feel the tradeoffs were worth the increased FOV, and that with current technology we have found the right balance. In the future when new technology changes the tradeoff equations we will find a new balance.

https://www.reddit.com/r/HPReverb/comments/kxczef/hp_checking_in_here_to_answer_questions/gj9hnzp/?utm_source=reddit&utm_medium=web2x&context=3

In other words because the G2 is using smaller panels than the Index (regardless of the FOV question), they have to use optics with a higher power, which I understand as magnifying more and in this case, it might be the reason the facemask is more recessed than the G1 because otherwise it would be too much lens curvature and aberrations. Having said this I still don’t get with the G1 with the same panels didn’t cause me any eye strain but the G2 with the same panel size is causing it to me. Either I really have a G2 I have to RMA, or more likely given the number of people complaining, they changed the optics focal length to be shorter than G1/Index/Others because they’ve chosen a Valve lens which material is producing less artefact and CA but conversely is refracting less as well. I can’t help thinking they’ve cut some corners with this to the point I’m starting to regret having returned my G1 when they’ve announced the G2, even more so that the latest WMR is correcting the G1 CA…

In thinking about the post I made in another thread, I decided maybe I should give my prescription lenses with progressive bifocals another try, just concentrate on central vision and ignore any blurriness in the lower part of the screen caused by the progressives. In real life, the central optics in my glasses corrects my distance vision to about 20/15 and repairs mild astigmatism. And in learning to move my head to get the sharpest part of the image more or less centered, I may even be catching a little of the +OD progressive correction to help me see the sharpest. Even small text in MSFS now looks pretty readable to me through the G2.

A corollary thought is that folks who haven’t worn glasses for vanity or because of the expense (“my vision is not that bad”) might want to get their eyes checked and get prescriptive correction if needed to milk that extra little bit of sharpness out of their VR headset. After getting a new eye exam, I still plan to get lens inserts for my G2 if for no other reason that it’s a pain to don and remove the G2 wearing my glasses.

a quick follow up with the Paolo +2 glasses: they are super comfortable to me compared to the other one I tried before (the Alden +1). The correction is a little bit too strong but didn’t cause any eye strain in particular. I’ll have to wait for an eye exam because of the pandemic but in the meantime I’ll probably try finding a Paolo +1.5 and this might be just what I need with my G2.

@JALxml the Paolo are also very easy with the G2 thanks to their fine branches. This make it easy to put and remove the G2 over, and the added benefit of the +2 is that I can use them both in the headset and when removing it (I just lean a little farter from my monitor than when I use my reading glasses). This alone is a huge pain removal, but I have to be careful because the Paolo frame/branch connection seems a little fragile.

I have a hunch there’s a software bug that causes alot of the blurryness.

Every single VR hmd uses CA filters to battle edge blurring. This CA filter must be centered to the hardware lens IPD or it will be applied incorrectly. I did some testing and I have a feeling that the software IPD for CA filter is applied incorrectly causing edge blurriness.

Steps for reproducing findings:
-Use Reverb G2
-Remove headset cover
-close one eye (your preference)
-Look at the lense at close range
1)
-Set the hardware IPD to the maximum setting. Go to the sweet spot. Start moving your eye towards the other lense. Observe the location of the perceived blurriness change compared to the lens edges.
2)
-Set the hardware IPD to the minimum setting. Go to the sweet spot. Start moving your eye towards the other lense. Observe the location of the perceived blurriness change compared to the lens edges.

Expected result:
The location when the blurrying starts to occur should not change if CA filter is applied correctly to the hmd

Observed result:
Blurrying starts to occur much more sooner when the hardware IPD is set to minimum location being in accordance with the expectation if the CA filter is applied to the hmd with a software IPD much larger than the hardware IPD.

This is more of an “can you see what I am seeing?” thing rather than a real bug issue as I cannot be 100% sure of my findings. I hope I’ve given good enough reproduction steps to confirm or deny my findings.

I was hinting this could be the newest WMR anti-CA causing the blurriness as well, because when I see a copy of a purple EFIS text displaying in blue half letter size above, it is not CA but anti-CA software to me. And like you’re saying, if the filter is not aligned and calibrated for both position (L/R/U/D) and focal (eye/lens distance) it can’t work.

If I understand what you’re saying, the anti-CA center point is fixed and doesn’t move with the IPD adjustment? This might be a reason indeed.

If you want to try out some (I tried with no particular success), there are hidden registry settings you can adjust…

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm\ExtendedComposition\ColorDistortion
REG_DWORD

1 = enable | 0 = disable

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm\ExtendedComposition\ColorDistortionB
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm\ExtendedComposition\ColorDistortionG
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm\ExtendedComposition\ColorDistortionR
REG_DWORD

1000 = 1.0 | 500 = 0.5 | 1500 = 1.5

NB: WMR reloads the values with the sensor! You can therefore remove headset, change value, put back headset and you can see immediate results!


I’ve otherwise set this one to 1 thinking it might delay loading WMR portal assets if WMR gets started by an application. I don’t know if this helps at all but if you do find it is giving you better start times or more VRam in FS2020, please share you experience back in this discussion!

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Dwm\ExtendedComposition\SkipBootArtWaitForVisibleRenderables
REG_DWORD

There is also this one I didn’t try yet but which makes me wondering now that SteamVR 1.16.1 can project up to 6 frames, or whether this could help reducing the apparent lag when moving the head:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Spectrum\HeadPredictionAmountFrames
REG_DWORD
2 by default on my system

Absolutely in my observation as well!
If you have XP11 there is significantly less blurryness.

Is the G2 still worth to buy?

In my opinion, yes, if you have the hardware to push it. I do also believe that these issues are partially software related and hopefully will be fixed sooner than later. Now it’s the state of VR, it’s not perfect, it’s crawling to maturity and it will take some time.

1 Like

I think I have found an easier way to look for the incorrect anti-CA filter in action.

Launch OpenXR developer tools and OpenXR demo scene. Look at the green bar. If you look at it at the correct orientation you will see a purple (red and blue makes purple) aura and a clear cut purple line. As a sanity (or insanity? :P) check tilt your head 90 degrees and look at the bar again. Purple haze and purple line are now gone. You can now tilt your head even more and when looking up side down the puprle haze and purple line should reappear.

1 Like

So if I understand the anti-CA IPD vs. the overall optical IPD, this might explain why some folks think the G2 is great and others can never get enough sharpness around the sweet spot. If the central optical IPD is 64 mm +/- 4 mm and the anti-CA IPD was designed to be at ~64 mm, perhaps if your IPD is right in the middle, you’ll think the G2 sharpness is great whereas if you’re off to one side, you won’t be so happy even with physical IPD adjustment if the anti-CA optimum is fixed right in the middle distance.

My best IPD measurement with the free iOS EyeMeasure app that requires a recent iPhone with RealDepth face recognition abilities is 63.7 mm and I set my physical IPD to 64.4, as the slight increase seems to help. And I now think, wearing my normal prescription glasses and ignoring the off-center progressive bifocal part, that central vision sharpness is great for a VR headset.

So it would be interesting to hear from folks who have IPD’s at the extreme what they think of G2 central vision sharpness. Maybe just as when folks talk about graphical performance in MSFS and give their rig specs, when folks remark on some VR headset sharpness, they should give their IPD, whether wearing glasses, contacts, reading glasses, or using unaided vision, etc. Otherwise, a declaration of good visuals or bad is like talking about poor FPS in MSFS without revealing the hardware behind it! :slightly_smiling_face:

2 Likes

For anyone who thinks a little +OD presbyopia correction might help improve VR headset sharpness, I found the following reading glasses on Amazon. They’re relatively inexpensive, only $8 a pair, and come in a variety of +OD corrections, including very low +OD values, increasing in 0.25 increments from a low value of +0.5, i.e., 0.5, 0.75, 1, 1.25, etc. (curiously, the lowest values show up in the middle of the “size” dropdown). Also, these lenses do not have anti-blue light blocking, which for some El Cheapo reading glasses on Amazon makes the lenses yield yellow-tinted vision. Since I’m not sure that a little +OD correction won’t help, I ordered +0.5 and +1.0 OD lenses and currently got $1.20 off the price of each item for a total price with tax of about $14.75 and, with Amazon Prime, the glasses will be here Tuesday, 1/19, to help me test where things stand relative to completely unaided vision (+1.5 and +2.0 OD definitely don’t help me with a G2). These glasses have ~the same dimensions as my normal glasses (see product images) but the IPD(or PD) is not provided in Amazon product info. Normally, I require +2.5 OD to read at 40 cm and the +OD correction requires falls off with increasing viewing distance but the rationale for trying some +OD, if you need extreme presbyopia correction, is in the following table: PSA: Reverb G2 small sweet spots, observations and solutions - #20 by JALxml

Even if the optical quality of these inexpensive glasses is not the best, they should at least enable me to test in a crude way whether even minimal +OD correction is helpful when viewing a virtual image displayed at some unknown distance of a few meters in a VR headset.

Amazon.com: Eyekepper Classic Vintage Style Reading Glasses: Clothing

Edit_Update: Actually, the product specs have the following info on blue light blocking and in the same part of the blurb (unquoted) claim good anti-CA properties for the lenses used:

We are improving our products based on customers’ feedback. The color of the BB40 lenses is lighter than the previous yellow lenses, result in avoiding color distortion to a greater extent, with 40% blue light blocking for wavelength 395nm-500nm.

1 Like

I can’t wait your feedback, thank you for helping contributing in this discussion!

Speaking of lens quality, the Paolo is a little bit less good than the Alden, the former having a little bit of faint ghost image for anything white on black but really faint, whereas the later is free of any. This might be also due to the +2 vs +1 correction though and the increased thickness.

1 Like

As a matter of fact there are some questions about G2 vs Index from time to time and I believe I’m nailing some of the key differences still making the Index a better headset to me besides LH tracking:

  • I can measure close enough HFOV: ±98º for the G2 and ±105º for the Index, but the G2 still makes me feel like looking through binoculars. Yesterday I’ve realized in fact the Index outermost FOV edges are not surrounded by as thick a black edge as the G2. I attribute this to the canted displays allowing them to be displaying closer and reducing the surrounding, effectively giving the impression of a wider open view. WIth the added much higher VFOV of the Index where the top edge is nearly non-visible, it really gives a better immersion into the VR world views.

  • The more I think and analyze this, the more I believe the G2 lenses are calibrated for a shorter focal length and they had to make this compromise in order to support the different refraction indice of the Valve lens material with the same small sized panels as the G1. However this is very uncomfortable to me. The Index focal distance is more than 2m and I can see clearly without any reading glasses with it, with no discomfort nor eye strain after many hours.

If it turns out the G2 focal length is shorter than other headsets (given the number of people complaining about the “sweet spot” there must be something there) I believe this is even more important because it can have further implications with the vergence/accommodation problem: Vergence-Accommodation Conflict

Yesterday I’ve tried pushing the Index even further on the same hardware, thanks to the new SteamVR 1.16.1 beta, and I must say running the Index at SS250 or SS300 (thus using TAA around 60/70 max) was giving me so nice a visual it was seriously competing with the G2 (even more so when you factor in the quality of the Index and the larger FOV). I know it sounds weird but despite the lower panel resolution, when pushing the Index at this level it was not much different in the end that my G2 in practice, because I can’t run the G2 at 100+100 (which is 3K x 3K). And the best of all, it felt natural for my vision to look around both far and close, and after a couple hours I was not feeling any strain either, despite pushing the Index to 160% luminosity sometimes (gorgeous when flying over Dubaii at noon!)

Out of curiosity: what resolution are you running your G2 at?

It depends!

See Balancing Performance and Visuals:
My 2070 SUPER VR settings and suggestion (Reverb G2 - WMR)

Ah, 2070S. I had an Index at the time I had a 2070S and sold it off as I couldn’t properly drive it (and needed the money). I regret that now.

Some further testing: With 300% supersampling and in the WMR treehouse I get vertical clarity from 12 o’clock to 6 o’clock but horizontal clarity stays as poor as it used to be. The same text that’s clear down low 6 o’clock is unreadable if I rotate my head so that the text is at the 9 o’clock position.

I’d take that vertical clarity over as a horizontal clarity any day than vice versa.

1 Like