Phr00t
-
Posts
47 -
Joined
-
Last visited
Content Type
Profiles
Forums
Gallery
Blogs
Events
Store
Downloads
Posts posted by Phr00t
-
-
In regards to wandering controllers in the virtual world while being held completely still in the real world: my battery life meters show full. My Odyssey+ controllers in the same lighting conditions are rock solid. It is possible a more ideal lighting situation would improve controller tracking, but we have to ask ourselves... how ideal does the lighting conditions have to be for acceptable Cosmos tracking, when significantly cheaper VR systems handle the same lighting conditions far better with fewer cameras? I'm hopeful HTC is continuing to improve in this area, because they need to. Glad it is working acceptably for others, though.
-
1
-
-
Just got done testing v1.0.7.1.
Tracking has been improving in v1.0.5, v1.0.7 and v1.0.7.1, but I wouldn't consider it acceptable yet. One thing not listed in your "controllers" section that contributes to perceived "bad tracking" is how much they move in the virtual world when held completely still. If I hold my controllers as still as possible, the controllers will slowly wander around the general area.
Getting the scene to render to the horizontal ends of the LCD panel would maximize the field of view possible, so I'll keep watching for at least an option to do so.
Ultimately, I have extended my return window by getting a used Cosmos because I like what I've been seeing so far. However, if after this new return window nears its end and tracking and field of view is better on my Odyssey+... I'd really hate to give up on it, but I think I'd have to. The hardware seems so solid and was really what I've been looking for.
If we can get those things cleared up, I'd keep it, share why I think it was worth keeping, and then I'd love to talk reprojection. 🙂
-
7 hours ago, TomCgcmfc said:
I'm glad you have noticed a FOV improvement. I cannot see this but I'll take your word on it that. It looks the same to me.
So, have you stuck with 1.0.7 for the better FOV or gone back to 1.0.5? Thanks.
The facial padding and flip-up design makes it difficult to maximize the field of view. Try pressing the headset to your face to get your eyes closer to the lens.
I'm sticking with v1.0.7 for now. Both the tracking and field of view seem better to me. Looking forward to more improvements, though.
-
This update is definitely a step in the right direction. There is a noticeable improvement in tracking and field of view (which I report in more detail here). More improvement could be made in both areas, although tracking has further to go.
I think the biggest problem with tracking, at least for me, was controller obstruction. They wiggle and get stuck quite a bit when crossing paths, but it is better than v1.0.5 was. Headset tracking still had a little wobble to it.
Anyway, much more hopeful this thing can be fixed and looking forward to more updates. I'm keeping the Cosmos for a little while longer.
-
2
-
-
I like the improvement made here!
Funny enough, my original FOVTester broke, so I actually had to write another one really quick (I'll get it up on github, full source, in the next day or two). Screenshot above, where you can see the measurement taken of the crosshair just getting lost in the vignette to the left.
My Odyssey+ can get the crosshair to about 50 or 51. I think we can get the Cosmos to 49 or 50 (which would give us another 2-4 degrees total horizontal FOV) by simply disabling the vignette altogether.
This is a perfect example on how you can increase the field of view of a headset with simple software updates! I hope this motivates your team to squeeze just a little more out here.
-
3
-
-
Yay, updates! Thank you for following up. Looking forward to trying these updates out in a few hours.
However, "Optimized the camera passthrough vignetting to better align and render safety boundaries" sounds like vignette changes were only made to the passthrough camera. Perhaps the issue was misunderstood -- the passthrough camera mode reveals how much more there is to see, and how much more field of view is possible with the device while in a game. If the vignette is pushed back in passthrough mode, it should be pushed back in SteamVR too.
-
I see there is a checkbox in the Vive Console for "Motion Compensation". I'd like to know a little more about the method the Cosmos uses, since I presume it isn't using the SteamVR compositor or its methods.
SteamVR has "motion smoothing" and "reprojection". Windows Mixed Reality has motion vector reprojection. Oculus has asynchronous time and space warp. There is a good article here:
https://uploadvr.com/reprojection-explained/
... that explains the different methods. A great image at the bottom summarizes the technology. Where does the Vive Cosmos lie on this image?
Oculus and Windows Mixed Reality are known for great reprojection. I've noticed quite a fair number of artifacts using the Vive Cosmos here. I know your engineers plates are quite full, but is this an area being looked at for improvement?
-
13 hours ago, TomCgcmfc said:
Sorry mate but your so-called tool (using java 8 and steamVR resources) looks like something you created ~4 years ago and you seem to be the only one using it. I wonder why? Maybe you are being a bit trollish? Hopefully not. Nor does it appear you have gotten any positive response to this on any other forum (like Steam) you have posted this on.
I also do not have much faith in your passthrough vs headset border views conclusions. Slightly increasing the FOV in the outermost part of the lens where the image is already well out of its sweet spot, and hence already blurry, does not sound like this is going to improve overall immersion imho. Still, might be something in it, but I doubt it.
Also, I believe that the published FOV for all VR headsets is diagonal, not horiz/vert. Plus, there are so many individual factors that ultimately effect each person's perceived FOV such as ipd, eye/lens distance, face shape, glasses wearer, just to name a few. So, ultimately I guess it is pretty subjective, and always will be for VR headset consumers. Given the similar sizes of all major PCVR headsets lens and geometry it is unlikely that FOV is going to vary significantly right now, and will likely remain subjective for some time to come. The Index (which is still a Valve experiment right now imho) uses a combination of adjustable eye relief (Rift S also does this) and 5 deg canted lens to potentially provide 130 deg (diagonal) FOV. Unfortunately, not all software has been modified yet to support their non-parallel (canted) lens.
Look mate, this is an open forum so please feel free to beat your little drum on it. Just don't expect everyone (including developers) to jump in and march to its beat, lol!
If you want to read more about the tool and its reception years ago, see the reddit announcement of it. People did their best to nitpick it, but it was also well defended. It is really just a simple tool, and there are other old ones that would probably give similar results (like this one).
If you want to read more about increasing the field of view by adding to the outermost areas, see how Microsoft researchers did just that and reduced motion sickness.
Many factors go into field of view, yes. However, the Vive Cosmos could remove one of those significant limitations (LCD utilization) to their field of view with a software tweak or option here.
I don't expect developers to march along, but I also didn't expect being called a troll for bringing up, documenting and demonstrating this issue.
-
1
-
1
-
-
24 minutes ago, TomCgcmfc said:
I realise that this this FOV issue is important to you but I feel the tracking issues and controller compatibility are much more important to most. Personally I think the FOV is fine. It is at least as good, maybe even a little better than my Oculus Rift cv1. I had a friend who has a Rift S over the other day and he thought it was better than his. From my very basic understanding the potential problem with trying to squeeze every last micro bit of border coverage is that you may end up with a rainbow edge effect. Probably not worth it for a couple of degrees higher FOV. In any case I'm confident that Vive will review this, just in case it can make a significant difference.
It is clear by the status update above that tracking issues are being worked on, so I wouldn't be worried that is being overlooked.
As I said in the other thread, you need to actually measure the field of view with a tool instead of relying on subjective "looks fine to me" sentiments. Oculus headsets are not known for impressive fields of view, so subjectively comparing to those isn't saying much.
This "rainbow edge" you are concerned about is visible in the passthrough side of the screenshot I provided (seen here). It is clear the rainbow effect is extremely minor, and considering it is just on your far periphery, it will be almost impossible to notice let alone be bothered by it. However, significant encroaching blackness on your periphery is noticeable and leads to that well known binocular effect.
At the very least, it should be an option to increase the field of view (by removing the vignette to the passthrough borders), because otherwise it is very noticeable comparing to my Odyssey+ (both subjectively and measured).
-
Thank you for the update and timetable. My return window is early November, so I should be able to evaluate these changes. I will be optimistic that "optimizations and improvements not mentioned here" refer to the field of view issues, which I'm looking forward to just as much as tracking improvements.
-
1
-
-
PLEASE share this image with your developers, it proves and explains everything:
-
You really need to use a tool to quantify field of view, and I did. You can use the linked tool above to do your own test on all SteamVR headsets. Turn your head until the crosshair on the floating box ahead disappears. Do it for all directions and measure the difference. 95 can "seem fine" but is disappointing if you are suppose to be getting 110, or at least results from cheaper headsets. Simply using more of the LCD panel could be an effective solution.
-
There is some hope here! Looking towards the right of the right lens, and left of the left lens appears to show a section backlit by the LCD panel, but NOT rendering anything due to the vignette. If this vignette could be disabled, it should offer a significant jump to the field of view.
-
After testing the Vive Cosmos, I can confidently say the field of view is much lower than 110 degrees, and very disappointingly so. My Samsung Odyssey+ puts up consistently better numbers.
I'm using the following tool: https://github.com/phr00t/hmd-fov-tester
The Odyssey+ gets me about 104 degrees horizontal, while the Vive Cosmos gets about 93 degrees horizontal. This is very noticeable in games. Vertical field of view was very similar between the two headsets.
I'm really hoping this is due to the much stronger horizontal edge vignette on the Vive Cosmos, and not due to optics or physical display. Please, please let us reduce this vignette effect, so the display can be fully utilized and hopefully realize the full field of view on this headset.
I'm happy to hear tracking is being worked on (hopefully also headset tracking). However, I haven't heard anything addressing the field of view or aggressive vignette. I did submit an issue via Vive Console.
-
Any expected time frame for next update (not that I want sloppy updates rushed)?
-
This appears to be the biggest problem after the lighting issue, which has seen some improvements.
What I find odd is, the hidden hand stops and jump around when occluded, even though the accelerometers shouldn't register such a jump or pause. Sounds like they need to trust the accelerometers a bit more when briefly occluded.
-
Even better: an adapter that says it supports DisplayPort 1.4 and the Rift S: https://www.amazon.com/Cable-Matters-DisplayPort-Aluminum-Compatible/dp/B07F17ZHJY/
-
It appears DisplayPort 1.2 runs @ 17.28GBPS, so you will need a cable or adapter that supports at least that speed. Many only support 5.4GBPS, which is the likely cause of this problem. I haven't confirmed this and still hoping technical support can chime in. This is the cable I have coming in tomorrow, which will hopefully work:
https://www.amazon.com/gp/product/B07H8BHZPN/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1
-
I'm having trouble using my Vive Cosmos on my laptop. My laptop has two DisplayPort connectors. One is a DisplayPort 1.2 connector that goes to the Intel HD card (since Cosmos setup complains that I need to plug it into the *other* DisplayPort for my NVIDIA card when trying to use it). My other DisplayPort connector is actually a USB-C port with an Alt Mode DisplayPort 1.3. It has a DP icon on the USB-C port, confirming it supports video output. I got a USB-C to DisplayPort adapter, which I plugged the Cosmos link box cable into.
Cosmos then gives me an error, 201: Headset display not detected. I was able to update the firmware, and it does say it tries to restart the device... but I keep getting a 201 headset display not connected error, telling me to check my cables.
I'm using the latest NVIDIA drivers for my RTX 2060. Please help and thank you!
-
There has been quite a bit of speculation on the types of cameras used on the Cosmos. Windows Mixed Reality and Oculus headsets reportedly use IR cameras, while the Cosmos uses RGB cameras. IR cameras are known for better low light performance. Many report poor exposure and dynamic range with the Cosmos RGB cameras while using them as a passthrough, which may explain why the headset has so much trouble with lighting conditions. If the root problem is the type of cameras used, then this can't be solved by software fixes, correct? Is there anything the team can say to assuage worries that core problems with Cosmos tracking can be resolved via software fixes, specifically addressing the types of cameras used on the Cosmos?
Field of View: Not getting close to 110, vignette problems?
in VIVE Console Software
Posted
Another interesting tidbit: the collision bounds ignore the vignette, so you can see them beyond the game scene. The scene should go right up to where the collision bounds render.