PSA: Reverb G2 small sweet spots, observations and solutions

User Support Hub Virtual Reality (VR)
61

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.

62

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.

63

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

64

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

65

Is the G2 still worth to buy?

66

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
67

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
68

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
69

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
70

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
71

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!)

72

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

73

It depends!

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

74

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.

75

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
76

You know, I complain a lot about immature people on this forum constantly whining and moaning about insignificant problems, and there is way too much of that.

However, then I run across threads like this (and @CptLucky8’s VR setting threads) that are full of so many people that are so smart, that I feel like a completely retarded moron in comparison. So, a grand thank you to those who meet that description.

What I’m taking from this thread is that I should probably get my eyes checked (which I haven’t done since I got LASIK in the early 2000’s), to see if I need corrective lenses, which I’ll probably order from here, and even if I don’t need corrective lenses, I may get the script-free version just to protect my lenses. My other conclusion is that I need some mod to the face gasket like this, which I’ll probably order and/or make as well.

But I’m not big on DIY stuff, so if I can order it with a single SKU from a single source at a reasonable price, that is absolutely what I’ll do.

So thanks for the information that is over my head, and also for restoring my faith that this forum isn’t just full of teenage crybabies. I’m fairly certain that those two optical solutions I plan to implement, plus @CptLucky8’s VR settings, will put me in as good a position as is humanly possible to get the max I can out of the VR experience in MSFS until software fixes and optimizations improve things that are out of all of our control, and of course, ultimately CPU and GPU upgrades just increase the horsepower of my machine on a HW level.

With a 10600K overclocked to all-core 4.8GHz, and a 2070 Super overclocked by a +175/+1000 bump, I don’t foresee much HW help in the short term. Maybe in a year or so when the 3080ti and/or Super models hit general availability, or even the 40-series, and maybe Intel’s 12th Gen (though I’m going to give their 11th Gen a good hard look) or perhaps Zen 4 from AMD, then it will be enough of a bump to make the upgrades worthwhile. We’ll see.

Sorry for writing a novel and going off topic, but I do that. A lot.

3 Likes
77

For those of us ( mostly all ) that need to protect the lenses of our precious HMDs, I always recommend the protective film from KlearKare.com (44mm circular version cut to proper shape) that sell for $8USD. It definitely seems like a “scary” process - spray soapy water on your fingers and apply the film to your lenses and squeegee the water out while hoping that you don’t trap many air bubbles or any lint. I have done this with moderate caution on my Odyssey+, my G1, my G2, and my Quest2 without it ever creating a real problem, although the extremely clear result might not show up for a weak even though I actually could use the HMD within hours even with a little smudging. The nanoparticles “heal” over time and it’s nice to be able to relax about possibly scratching those all important lenses and/or your own glasses. Check out their website although their demo video is for a flat surface (a cellphone) instead of our complex curved lenses. Make sure you order the 44mm which seems to be just larger than my HMD lenses. SKU:KK-WATCH-44MM

BTW, I tested my G2 for a couple weeks and decided that for my face size and eye prescription, that I liked my G1 image better like several others on forums so I sold my G2 on ebay.

1 Like
78

Does anybody know where I can find the registry entry to disable the CA-filter?

I found it somewhere a few days ago but forgot to bookmark it…

79

Really weird. @CptLucky8 and I have gone completely in opposite directions on the possible need for +OD help for presbyopia. Initially based on fooling around with a smartphone VR headset and a kiddie Google Cardboard VR education science kit lab, I thought I’d need plenty of +OD correction - +2 OD seemed to help with the smartphone VR headset vs. no glasses. My normal prescription for reading at 40 cm is a +2.5 OD correction.

Now that I actually have a HP Reverb G2 headset, my conclusion is that no additional +OD correction helps my native unaided vision, the +0.5 OD reading glasses ordered from Amazon are worse, and +1 OD worse still. And trying unaided vs. my normal prescription lenses with graduated progressive bifocals again, the unaided vision is better because the increasingly progressive bifocals mess up the lower part of the visual field. My left unaided eye is definitely worse than the right so I think ultimately prescription inserts to correct both eyes individually, both for myopia/hyperopia and astigmatism, will work best.

Although I downloaded both the Steam free version of TestHMD and the paid version bought on the TestHMD itch.io website, since one has to run those under SteamVR, I decided standing out on the terrace of the Windows Mixed-Reality Portal Cliff House and looking back inside toward a display of Windows Store offerings was the best and easiest test for static visual acuity, as shown by the picture below. I moved the Store object to the distance shown and looked at the finest text on the popped-up Store offerings. For me, unaided viewing plus the physical IPD distance displayed in the Settings, Mixed Reality, Headset Display slider box gave the clearest view of the finest text on the Store pop-ups. I use the white roof support and the white background of the Windows Store logo as a test of chromatic aberration and didn’t notice any (maybe I’m visually “tone deaf?!”). Looking at the individual blades of grass on the green terrace lawn and also to the right and up into the sky to see the stars (or whatever they are) helped confirm that my viewing conditions were pretty good.

BTW, if anyone knows how to save a VR screen capture efficiently, I’d love to know. After calling up the Windows VR “Start” screen with the Windows button on the controller, taking a picture, getting a floating screen capture window, the only way I could figure out how to save the picture was from the title bar middle right sharing icon to share the picture to Snip & Sketch! Then save the pix from there.

CptLucky8 recommends in his settings advice to use low-resolution and 720p as the settings for the Windows Mixed-Reality Portal screen to have more resources for MSFS. For the purposes of such an eye test, if you otherwise use CptLucky8’s settings, remember to be sure you have your Settings, Mixed-Headset Display, Quality of My Home settings set to High for level of detail and quality of effect and 1080p for window resolution for the purpose of using the Cliff House as a visual acuity chart.

20210119 Cliff House Screen Capture01
20210119 Cliff House Screen Capture011920×1080 149 KB

P.S. The sharpness of the image in the G2 headset is quite a bit better than captured here in the picture and saved as a .JPG and then site-processed after upload.

80

This is really really interesting a report, thank you for your time with this. I’ll certainly try the same vantage point and confront what I do see with my G2 and I’ll not any single problematic detail to compare.

There could be another explanation as well: my G2 might be defective (calibration, lenses wrongly adjusted and sealed, etc…)

Can you tolerate the headset without any eye strain after 1h or 1h30 too?

1 Like