OpenHMD 0.3 Released With Support For More VR/AR Devices

[phoronix]

Phoronix: OpenHMD 0.3 Released With Support For More VR/AR Devices

Nearly three years after OpenHMD 0.2, the OpenHMD 0.3 release is now available for the open-source SDK and support for various AR/VR headsets...

http://www.phoronix.com/scan.php?page=news_item&px=OpenHMD-0.3-Released

[coder]

You forgot to mention they have exactly one device with positional tracking support, and it's one I'd never previously heard of (and probably therefore one that probably nobody owns).

Worse: for all their devices, no OSVR support.

[kpedersen]

Worse: for all their devices, no OSVR support.

In theory, I would agree but in practice, I realized that it is almost impossible to actually get hold of a HDK / OSVR headset so frankly I think it should be the least of the OpenHMD project's priorities to support vaporware hardware.

I have just started a new VR research project and I am so thankful that OpenHMD exists; the actual commercial offerings are absolute trash. I am especially disappointed in Valve that their OpenVR implementation (SteamVR) forces you to be under the same shitty DRM system as their stupid kiddie Steam games. Absolutely disgusted by that in fact; bunch of sodding clowns.

On the plus side, OpenHMD (and reverse engineering all the proprietary hardware as usual) looks to be the way forward, even the Khronos standardisation (OpenXR) looks to be using it as a backend.

Offtopic: Not sure why Khronos allows Valve or Facebook to be members of the OpenXR group after such a disservice they have both done to the industry. Even to this day they have provided absolutely no code or open documentation pertaining to advancing VR. Such an embarrasment.

[coder]

In theory, I would agree but in practice, I realized that it is almost impossible to actually get hold of a HDK / OSVR headset so frankly I think it should be the least of the OpenHMD project's priorities to support vaporware hardware.

The same could be said of many of the devices they support - are you really going to go out and buy a Rift DK1, DK2, or CV1? No, but many people still have them. And it wasn't vaporware - they were truly available, but (like a number of recent VR products) have since been discontinued.

I bought into OSVR (HDK2) and am still interested in using it, in spite of the fact that poor Sensics couldn't keep the doors open. I'm thankful that Monado does support it. I will eventually upgrade to something better, but I'm holding out for wireless. Plus, pretty much any upgrade would require a new PC, at this point.

Don't forget that OSVR was out there before practically anything else (besides Oculus' proprietary DK1). It was a noble effort, even if it fizzled.

I have just started a new VR research project and I am so thankful that OpenHMD exists;

Did you check out Monado? If you want something that's actually standards-based, that's the way to go. As you noted, for most devices, it still uses OpenHMD, underneath.

even the Khronos standardisation (OpenXR) looks to be using it as a backend.

Not the actual standard, but one implementation thereof.

Off topic: Not sure why Khronos allows Valve or Facebook to be members of the OpenXR group after such a disservice they have both done to the industry. Even to this day they have provided absolutely no code or open documentation pertaining to advancing VR. Such an embarrasment.

First, I think Khronos' charter would need to be updated to specify clear grounds upon which to deny membership. As it stands, they probably have to accept anyone willing to pay the dues and agree to their terms of membership.

Second, the fact that these guys are members might actually be them trying to do the right thing. Let's not punish them for that, unless you have any evidence that they're playing a counterproductive role, in OpenXR.

[kpedersen]

Not the actual standard, but one implementation thereof.

Yep, I was referring to the implementations mentioned on the OpenXR web page.

• Microsoft: OpenXR runtime for Windows Mixed Reality headsets.
• Collabora: Open source OpenXR implementation for driving OpenHMD-compatible headsets on Linux.

Monado is actually the project that pointed me towards OpenHMD. It had a company backing it (Collabora) which in this day and age, ironically scares me. Either the project will die pretty soon, they will only support proprietary platforms and Ubuntu or they are more likely to bugger up and start making "lite" versions whilst keeping much of the functionality proprietary!

That said, I am possibly bing overly harsh. I am very glad Collabora is developing an OpenXR implementation. However, I prefer to keep to FreeBSD so I have had to look elsewhere.

First, I think Khronos' charter would need to be updated to specify clear grounds upon which to deny membership. As it stands, they probably have to accept anyone willing to pay the dues and agree to their terms of membership.

Second, the fact that these guys are members might actually be them trying to do the right thing. Let's not punish them for that, unless you have any evidence that they're playing a counterproductive role, in OpenXR.

Hmm... well at least they are having to pay (indirectly funding an open platform). However most of the work I can see is done by open-source developers and reverse engineers; companies like this are more of a hinderance and then will very likely steal the limelight once OpenXR is released and sell themselves as the pioneers who first understood the importance of democratisizing VR development. Bunch of phoneys.

[CandyAngel]

Their 0.3 launch blog post states

OpenHMD has a ‘usable for everyone‘ policy on releases

which is demonstrably not true. The NOLO driver is known to not work with a specific firmware. No effort to find out why and fix (if possible).

Justification is "only the latest firmware is really supported and you should upgrade", but you can only upgrade the NOLO firmware if you have Windows. People who don't (or are unwilling to) have Windows, however, fall under the category of "everyone". Thus, the "works for everyone" policy is a huge misnomer.

I was on the OpenHMD team until recently (this incident being the tipping point for me to leave) and specifically didn't change the firmware on the NOLO I purchased in case it had a different firmware. It was extremely disappointing to find out that even though I had forgone use of my NOLO for months (putting on hold development of several projects that require positional tracking) and doing so had paid off (discovering the non-working firmware), absolutely zero effort was expended to find out why my NOLO didn't work.

Unless you count telling me, someone who doesn't code in C and is totally unfamiliar with the NOLO packet format (unlike the person who said this to me), "you can try and get it working before the release in a week or so" as effort, of course.

tl;dr team member discovers driver doesn't "work for everyone", so doesn't meet requirements for inclusion, driver gets included in release anyway