Announcement

Collapse
No announcement yet.

Khronos Officially Releases OpenXR 1.0

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Khronos Officially Releases OpenXR 1.0

    Phoronix: Khronos Officially Releases OpenXR 1.0

    Back during the Game Developers Conference was the release of the OpenXR provisional specification by The Khronos Group while today for SIGGRAPH they have formally announced OpenXR 1.0...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    It kinda feels like VR was a fad and has already died off. Is there still the same level of interest in it?

    Comment


    • #3
      Originally posted by FireBurn View Post
      It kinda feels like VR was a fad and has already died off. Is there still the same level of interest in it?
      For me there is. I just want it to get more mature with better price points before I start buying in. When Navi\RTX level of performance is in the $200 range and quality headsets are as well, that's when I'll start considering it. It just isn't worth it if you can't pair good stuff together and you really need good stuff to push quality graphics to two high resolution screens else you risk VR sickness and headaches and whatnot.

      There's also a very big military interest in VR technology. A drone operator, for example, would be able to replace a multi-monitor setup with a limited view to a VR headset that can see anything, anywhere just by looking around, pilots or tank operators could use it as a way to have no blind spots, soldiers on the ground could use it to overlay an image to glasses to highlight potential targets by showing where it's detecting heat sources, movement, etc. Then there's commercial applications like training simulators, safer welding hoods, all the same pilot/drone/bind spot stuff, search and rescue could use the same overlay soldiers use. And then we have perverts which we'll leave at that.

      VR is not going away.

      Comment


      • #4
        Originally posted by skeevy420 View Post
        VR is not going away.
        It may be simmering on the back burner for a while yet though...

        Comment


        • #5
          For me the libODK was lame and only really supported Windows on newer devices. Shame because the API was actually quite nice.
          The OpenVR from Valve looked great but the implementation (SteamVR) required frigging kiddie Steam and its daft DRM so was a no go.

          OpenHMD and reverse engineered drivers looks good and I believe OpenXR implementation Monado uses it underneath on Linux.

          Whats great is that OpenXR and OpenHMD does away with all the consumer lies so you don't need a high range GPU. You can even run it on a Raspberry Pi (zero) (http://www.openhmd.net/index.php/showcase/vr-zero-pri/).
          Last edited by kpedersen; 29 July 2019, 10:57 AM.

          Comment


          • #6
            Originally posted by FireBurn View Post
            It kinda feels like VR was a fad and has already died off. Is there still the same level of interest in it?
            Not in consumer area.

            Interest is kinda dying off for businness and military use as well.

            Comment


            • #7
              Originally posted by skeevy420 View Post
              There's also a very big military interest in VR technology.
              Showing aggregated camera feeds in a head-mounted display is a lesser form of VR that was quite frankly possible also in the past.
              A drone operator, for example, would be able to replace a multi-monitor setup with a limited view to a VR headset that can see anything, anywhere just by looking around,
              debatable.
              For big drones (i.e. Predator) this isn't adding anything over a multiscreen setup, which is not a problem because these drones are controlled by people in a military base.

              For smaller drones this isn't adding much over a screen as the camera will have limited view anyway.

              pilots or tank operators could use it as a way to have no blind spots,
              Useful, but as said above that's kind of very very watered down "VR" that does not really require a lot of the fancy technology true VR needs.

              soldiers on the ground could use it to overlay an image to glasses to highlight potential targets by showing where it's detecting heat sources, movement, etc.
              This is called AR, Augmented Reality. And it's an entirely different kind of animal.
              The processing power required to effectively overlay something over real images requires significant processing power.

              Then there's commercial applications like training simulators,
              ok, but for a lot of things it's not real enough. It can't realistically simulate movement, a physical environment you can touch and interact with, or heat or a bunch of things that aren't just visual.

              safer welding hoods,
              wtf lol no. Camera-based visors are NOT VR, but a display and camera mounted on your head so it moves with you.

              all the same pilot/drone/bind spot stuff,
              civilian vehicles have much less visibility issues than tanks, because... you know.... they have large trasparent windows. Parking cameras are more than enough for trucks and other vehicles with blind spots.

              search and rescue could use the same overlay soldiers use.
              Same as above, this is AR. For civilian use you are not worrying about getting shot at, and there are better alternatives like say trained dogs.

              VR is not going away.
              It's not going anywhere.

              Comment


              • #8
                Originally posted by starshipeleven View Post
                That's a lot to quote
                Yeah, I'm aware I'm interchanging VR and AR because the underlying technologies are similar enough and intertwined in a lot of the use cases I'm imagining. The ability to take a video feed and then take the data from drones, people, aircraft, ground vehicles all into a virtual environment at the base for commanders to use or an overlay over what can be seen is still going to be a major factor in pushing this technology forward.

                I know that they can see quite a bit with multi-monitor setups, but you know as well as I do that using controls to move the view around is slower than naturally moving one's head and looking around. Think future forward to remote operated fighter jets operating a mach 3, pulling 15 Gs, and outperforming anything we know of that's declassified, the person operating that is going to want to be able to look and see with the minimal latency possible and a spherical camera setup and headset will provide that. Not to mention having to operate camera view controls as well as flight controls all at once. For a Predator styled drone the multi-monitor setup works; for a hypothetical fighter/interceptor drone they don't make much sense at all outside of cruising to the target.

                Comment


                • #9
                  Originally posted by kpedersen View Post
                  [...]Whats great is that OpenXR and OpenHMD does away with all the consumer lies so you don't need a high range GPU. You can even run it on a Raspberry Pi (zero) (http://www.openhmd.net/index.php/showcase/vr-zero-pri/).
                  This. I've ran some tests with OpenHMD on a old laptop with onboard Intel GPU, and it performed surprisingly well with the Rift.

                  Oculus is IMO not a platform worth developing for, unless you're strictly interested monetizing games (or apparently crappy & glitchy video players) via the Oculus/Facebook store platform. You can't just use the Rift for anything experimental/work/serious/professional application.
                  At the company I work for we use it for VR therapy/treatment/education, we we're forced to take the machine off the grid to prevent Oculus from routinely self-destructing the installation though updates, but also to prevent Windows 10 and Oculus (the app) from exfiltrating confidential medical (patient) data to Facebook and Microsoft. But disabling Windows update also violates company security standards.

                  So the only way to legally use VR in medical setting (in Europe at least) would be through auditable open-source drivers (OpenHMD/OpenXR), and on Linux (not Windows).

                  Oculus driver on PC is unstable, sometimes it will not startup properly, requiring reboot. Sometimes the HMD is not detected (yet reported as correctly working when the HDMI cable is unplugged), and the only way to fix it is to re-install the software. Which means downloading a 3-4 GB installer (of which 1.2 GB is the insanely bloated Oculus Home background app) over WiFi. Because we cannot trust to connect with a cable this machine. You have to repeat the download process a few times, because the installer usually corrupts a few files. And when you restart the installer, it deletes the existing download cache. Takes about a day typically. And when all of that has miraculously succeeded, you're then forced to sit through a 1 minute 'safety video' + 5-10 min calibration sequence (for controllers we never use).

                  The Oculus Store routinely pops up with horror/shooter/anime-rape-simulator splash screens, which looks kinda unprofessional during demonstrations or treatments for PTSD sufferers. Oculus Home constantly plays annoying idle sound loops in the background, which distracts patients and therapist alike when not using the HMD. I could go on and on.

                  In a nutshell the Oculus PC software is a non-stop train wreck, and undoubtedly the main reason why Oculus Store is doing poorly (which I assume it must be doing). It is an affront to all self-respecting PC/workplace/developer users. As a developer, I wouldn't even want to support any Oculus hardware until it's been fully de-Facebooked via open-source implementations. The Rift hardware is fine (in some degree even preferable to Vive, due to trivial & compact h/w setup), but the software makes it (including your PC) utterly unusable for anything but the bundled games.

                  SteamVR (an OpenVR implementation) is more mature than Oculus, and gives the user/developer much more control. But it is is weighted down by Steam. It is far more stable, gets less in the way, and Valve at least treats the PC user with honest, well intended respect (if you ignore Steam account/DRM). Our Vive on-site demo hardware rig takes 30 min to setup typically, and SteamVR calibration takes 1 minute or less (compared to the ~1 hour getting Oculus software under control). I did write a custom app to quickly test & debug the calibration for on-site demonstrations, as even the simplest Vive demo app is very slow to start and bloated.

                  And then there's OpenHMD. It's a few kb, compiled into the stand-alone application. No installation of driver required. It gets the Rift working in less than a second on Linux. 1 second software setup vs the usual 1-2 hour. Just plug in the HDMI cable, double click the .exe and you're up and running. Brilliant!
                  Of course, a calibration step will be required once position tracking is fully supported. Hopefully OpenXR will make calibration a one-time effort, with the settings shared between applications. And I bet Valve will fully support it.

                  On a side note, the Oculus Go is also interesting because of price and lack of cable. But it runs Android and is likewise locked down on all fronts. You need to have a mobile tethered via Bluetooth during setup for it to work at all (seriously!), so Facebook can also snoop on your phone (even though you don't otherwise need it). Like on the PC, Oculus gives no user control over the Go whatsoever. For example Facebook claims it has 70 Hz support and variable rate rendering, but you can't enable it, or otherwise determine if that is working at all. App development is largely undocumented and appears unsupported. Video streaming (display mirroring) is utterly broken, with no timeline for fix.
                  Root + some Linux distro + OpenHMD might make the Go useable for serious applications, but it is unlikely that anyone will spend (waste?) time on this device. Great hardware, shitty software. It's pretty much dead-on-arrival due to chicken vs egg syndrome.

                  To catch my drift. The biggest issue with VR is not cost, it is not hardware. I can survive the crappy VR UIs. It's not even the driver quality. No, it's the horrendously shitty bundled & forced software. Oculus/Facebook. Google. Microsoft. Valve, to much lesser degree. They are shitting in their own yard, it boggles the mind. It's like they want VR to fail, but want to be seen doing something with it.

                  So, looking forward what Khronos/OpenXR will bring on the software front in the (now) very near future. All OpenHMD have to do is complete the support matrix (i.e. mostly positional tracking + controllers). Once that's done, there's no reason a developer wants SteamVR/Oculus API (unless you *want* to be locked down to their stores for self-flagellation).
                  Last edited by Remdul; 29 July 2019, 06:29 PM.

                  Comment


                  • #10
                    Originally posted by FireBurn View Post
                    It kinda feels like VR was a fad and has already died off. Is there still the same level of interest in it?
                    According to the Steam Hardware Survey the number of VR users caught up with the number of Linux users in January 2018 and is further ahead now. Given the financial barriers to entry (expensive headset + high-end PC vs free OS and any PC) I think that indicates healthy interest.

                    There have been VR-only games at the top of the Steam best sellers list.

                    Comment

                    Working...
                    X