Calibration Veil compare method on CV1

[voidxno]

Created a simplistic method to compare the calibration/veil conditions on my CV1. Is a follow-up to my older forum thread (Two and a Half problems with my CV1). Like my god rays compare method (CV1 vs DK2) a few weeks back. It's an easy and quick way to see yourself how it looks on your CV1 (no need to take photos).

No Blender rendered 360 video created this time B) Just a quick and dirty constructed RGB gradient image used in Oculus 360 Photos app. Serves its purpose, and is easily reproduced in the future. Material and more thoughts further down. How it looked on my CV1:

Here are left and right photos shown together. You can start to see some of the differences:

A zoom/crop of the center makes it more obvious:

In the process of taking photos I found it possible to lock focus at a point visualizing some of the edge effects of the fresnel lenses. The zoom/crop photos below are NOT from photos above. These ridges are NOT visible under normal viewing. But it might have something to do with my veil/coarseness problem:

All photos have been taken under identical conditions and locked manual settings. Matches my previous illustration attempts and what I see myself alternating between closing left/right eyes. Though, as I've said before. The calibration/veil problem is not prevalent in most cases. Perception of it as a problem will also vary from person to person. But for me, when it's there, it destroys my immersion 😐

Thoughts on photos:

Reason for these photos is to illustrate and try to understand my biggest gripe with my CV1. In certain situations with large uniform light/dark sections. I get a visual effect of having a static dirty overlay veil between me and my VR world. Plastic protection on both lenses have been taken off, and have checked for possible residue.

Here is how I observe the differences in photos above. There is a red tint on right screen. They have different brightness/calibration levels. A veil/grain/coarseness overlay difference exists between them. Best illustrated on photos 4 and 5. I think this last issue is what contributes most to my problem. Those streaks on left eye photos are not motion effect.

After having taken these photos. Looking back on the problem and previous photos. I'm starting to wonder if the veil I'm seeing is the result of my fresnel lenses having markedly different finish.

One effect not visualized especially good in these new photos are the 'sprinkled stars' I can see on left eye with totally black background. It was touched on in my older thread with this photo:

Again. Starting to think these stars are light effects from ridges on fresnel lenses having different finish.

Just my layman non-pro take on it. Call it like I observe it. Not long before I go the support ticket route. Bracing myself for the support impact B)

Material to check yourself:

Those interested in checking yourself can download .zip file attached to post (VRCalibVeilTestPhoto360.zip). It contains a '_info.txt' that explains how to use Oculus 360 Photos. And a .png 4096x2048 image file that looks like this:

It is not a correctly formatted 360 environment image. A quickly constructed RGB gradient alternating between RGB values #30/48 and #0/0. The lines/bars looks ok in a 360 VR viewer because they are in the middle of the image.

I did create a video version of this pattern that rotated the RGB gradients horizontally around. Problem is that with 1080p recording I was not able to have enough resolution to capture the effects. So had to fall back to still photos.

Method used to take photos:

Taped my headset on the end of my table. Removed the facial foam insert. Is the closest I can get to CV1 lenses with my iPhone SE. While still getting 100% still photos, 90° and parallel to both left and right lens. Would have liked to be closer like I do if only capturing on right eye. Maybe try to get a good small camera module sensor solution in the future. The lego/camera contraption Doc-Ok uses is a brilliant engineering solution.

Prepared my iPhone SE with the Manual app. Locked everything I could: 5000K, focus-lock infinity, ISO500, 1/5s (fresnel focus photos in 4/5 had focus-lock set to 4th 'dot' from the right in the Manual app).

Got test image ready in CV1 through Oculus 360 Photos. Blanket over my head, and partly headset, to keep light out. Position iPhone left lens, take photo, right lens, take photo.

Equipment / Software:

Oculus Rift CV1, Win 8.1 x64, NVidia 980 Ti (361.91), Oculus Runtime 1.7.0.

[gamefan101]

Yeah, I get a cloth-like, veil effect in certain lighting conditions. I noticed it the first time I put the CV1 on. I put it down to OLED limitations.  Vive referred to a "moire"  and say they have corrected it..

[voidxno]

Quick update. Remembered that videos are just a series of still images. Timelapse experiment to the rescue. Not perfect, but closest I have gotten to illustrate the veil/grain/coarseness overlay issue. First part is just a rehash of my post above. Jump to 0:34s for new material):
https://www.youtube.com/watch?v=Ow65G4Uc8G8

Same process as still photo above. Just more of them, and a lot more convoluted B) Funny CV1 trivia: Did you know that CV1 sleeps OLED screens after 10min of inactivity on the gyros, even if head-sensor is active? And then spikes the brightness of screens 2-3min before down to normal? I know now 😐 Some flicker present in timelapse videos. Eliminated most with some timelapse/deflicker tools, without sacrificing details.

Finally got to try another CV1 at a friend of mine. Also brought mine to switch and double-check a few things. Basically his two OLED screens are nearly identically calibrated. Nowhere close to the differences in mine. My right eye screen is pretty close to his two. Except he even has a somewhat tighter black level on green stripe.

PS: Just saw the post about runtime 1.8.0 red tint update logic. When I get time I'll check out if it changes my CV1. If observable I'll try to post some comparison photos.

[mbze430]

If that RGB photo being display with the white bar is a true 255,255,255 then you can clearly see the red tint effect.

I actually have done this with a 20% gray display through the HMD, and can see it even better.  But great job on showing the rest of the "blinded" Oculus world.

I have seen this happen too in white scenes, but my red tint got fixed last night.

[Anonymous]

Eagerly awaiting the new Firmware to be tested by you.

[voidxno]

@mbze430: Thanks. Agree with you on veil effect. Was the intro/setup when I put on my CV1 for the first time I was wondering what the heck that grainy overlay effect was. A situation probably similar to your uniform 20% gray testing. Not exactly sure about the 255,255,255 on RGB photo. But it is a generated image, so RGB values are true/synthetic. Image/PNG then shown through Oculus 360 Photos. Highest RGB value on white bar is hex30/dec48(of255). That low because of strategy of photo exposure range to illustrate the effects. Sorry if I misunderstood 🙂

@manu: Then you're in luck B) I was still on runtime 1.7.0 today. Blocking Oculus in firewall to have some control. Did a quick view of RGB test image in CV1. To have a fresh reference. Upgraded to 1.8.0 with its newer drivers. New test of RGB test image in CV1. And yes, difference on certain parts, especially red tint/haze. Two new photos taken.

Here are comparison photos left/right on old runtime 1.7.0 and new 1.8.0 w/fix:

You can clearly see markedly difference and removal of red tint on right side (white area middle). Easier to see if we zoom in. Next two pictures also shows before (10) and after (11). Really recommend to right-click and save those two pictures locally. Then flip between them in a picture viewer. Would have liked to make an animated GIF. But 256 palette/dithering would kill color and details.

Before picture:

After picture:

I observe the following changes. Matching how I experienced it with my own eyes. On left side, not much have changed. But white part have had a tiny amount of red tint eliminated. On right side though. You clearly see the elimination of red tint/haze. I also think the different primary colors are much better calibrated on lower values. Especially green. Together it makes for a uniform stepping up from black level. Much more pleasant viewing. Also red color bar matches much better in its tint between left and right side.

Kudos to Oculus for getting this fix logic rolled out.

[voidxno]

Small update. Got thinking, and hobbled together an animated GIF of only the white area of before (10) and after (11) picture with 1.8.0 fix. The 256 palette works ok'ish. Here it is:

Also in the same vein. Created an animated GIF for panning motion veil effect from YouTube video:

Right eye in panning sample does not have the red tint fix obviously. Hoping to not repeat that convoluted timelapse process again. Veil effect on right eye is still there after the 1.8.0 fix. To be fair though, I think it is less pronounced without the red tint. Coarse left eye vein/grain overlay still going strong.

[Punisher]

I'm on my 3rd oculus rift now after getting 2 replacements, and they all suffered from this problem. A lot of terms keep going around: dirty window/dirty veil/grain etc. I know that 3 is a low sample size but to me it would seems like most rifts will have this problem to a certain degree.

The veil/pattern was different on every rift for me. In retrospect the first was the best, the one I have now the worst and the 2nd one had some additional defect that magnified the problem. I can't remember for the first rift, but my 2nd and 3rd rift only had it in the left eye.

One thing fixed it though: turning spud off. When I turned spud off on the 2nd and 3rd rift (never tried it on the first), the veil overlay disappeared. Of course it introduces the mura effect and the picture-quality is decreased. I wouldn't call it a solution to the problem. Also the veil/grain overlay might still be there being obscured by the mura, but it isn't layed over the experience like with spud on.

My guess is that there are problems with the rift: it's either that the production of the lenses is imperfect, the screens are not up to par or calibration is off. Because the overlay disappears with spud off I'm thinking that the lenses might not be the culprit here? Could it be that the screens' quality is insufficient to be run with Spud on?

I've messaged support that I'm not happy with my rift replacement, but they have declined to replace my rift once more. I have sent screenshots but the problem is just not evident in them. I guess a good video would expose the problem, but I haven't been able to take a good video...I don't think I have the hardware to be able to. They replaced my first rift on good faith when I showed them a video someone else made that resembled my problem. I totally understand the hesitation to send me a new one; I also don't have a lot of confidence that I will suddenly receive a rift that doesn't have problems. Don't get me wrong by the way: support has helped me very diligently and in a timely manner. I can't provide any proof that my headset is broken right now so I can't blame them for not replacing the headset once more.

It's frustrating to be told that it must be in your head though by support, and that most people would probably not find fault with the same headset. I wish they would acknowledge there is a problem here, I'm pretty certain that they must know that this is a big issue, comparable to the red tint issue. Because there hasn't been as much talk about it as red tint, I'm not sure if Oculus is working on a fix (if it's a calibration issue).

I want to thank Voidxno for this topic as its the best one about the veil/grain overlay problem. Have you tried running your tests with spud off? It might narrow down the cause.

[Lemny]

I have my fourth Rift now an all had this problem.
I made a post about it: https://forums.oculus.com/community/discussion/43813/does-anyone-have-a-flawless-oculus-rift#latest

[voidxno]

@Lemny: Thank you for sharing your experience and photos. It's a tough crowd though. You need to balance on a razors edge and dodge derailment bullets at the same time. Unfortunately. The drop-off to 0/black also puzzles me. Havn't tried to delve more into it for now. Saddens me that people do not take it more seriously. Like good PC monitor reviews (TFTCentral.co.uk). There is no reason VR headsets should not be held to the same standards. Calibration, color accuracy, black depth, gamma, white point, and more. It only enhances the VR experience I though we all strived for.

@Punisher: Thank you for a very detailed post. Good to compare experiences. For the sake of completion. Did a very quick session with spud off. Here are the results:
IMPORTANT(for others): Photos 13 and 14 below DO NOT represent standard CV1 operations! Corrections (spud) has been turned off through a DEBUG registry entry!

Zoom/crop:

When viewing it with my own eyes I agree that it reduces/changes the vein/grain overlay effect. Looks like the visibility is connected to how things are calibrated/tuned. Still present though. Just like you I cannot wrap my head around if the effect is connected to finish of fresnel lenses or some OLED screen coating. Or a combination. Can only speculate without more info from Oculus. Even if not much talk about it, I hope Oculus has some focus on it. One of the reasons I have waited so long before going the support route.