Galaxy XR: The first Android XR headset

Kind of sad to see here on "hacker news" that 80% of the comments are low effort cheap shots.

The interesting thing here is the core of it, being Android XR and its deep AI integration, especially the spatial awareness. Devices will come and go, but the OS will be the core that stays and grows and evolves over time. I am very curious to know how much of this is all exposed as OS foundations to build on vs a monolithic app built to look like an OS by Google. This has been a large part of Meta's mistake, where the OS is not providing many of these fundamentals and any app you see doing it is mostly re-inventing it themselves or relying on 3rd party tools like Unity to do the heavy lifting.

The really impressive part of Vision Pro is actually how well thought out the OS is underneath it, exposing fundamentals of how 3D computing can work. Especially the part to do with compositing together multiple spatial apps living together in the same shared space and even interacting with each other (eg: one app can emit a lighting effect that will shade the other's rendering).

I am very curious if Google has done this kind of foundational work. Especially if that is designed (as they claim) from the ground up to interface with AI models - eg: a 3D vision language model can reason across everything in your shared space including your pass through reality and respond to it. This would be truly amazing but there's zero technical information I can see at this point to know if Google really built these foundations or not here.

> one app can emit a lighting effect that will shade the other's rendering

I always felt this was such an outrageous burden to developers. Its cute and all but really, who cares? I don't need one desktop window to emit light on another window. Is that really worth having to remake or modify every asset?

That said, all the work they did around laundering click and gaze information for privacy was nice to see.

> Devices will come and go, but the OS will be the core that stays and grows and evolves over time.

Say that to my Google Cardboard SDK programs, or the Google VR SDK ones, or Google Daydream ones.

You couldn't have chosen a worse topic on which to dump a generic "ranting about Google abandoning projects is low effort cheap shot", because Google does abandon VR projects (including OSes and APIs, not just devices) every 5 years, almost like clockwork. What I would call "a cheap shot" is to think that this new fancy "OS" will be any different. In fact, I pity the people who still consider jumping on this particular bus _again_.

Why on earth _did_ they abandon cardboard? It was really good for getting VR in the hands of.. well, everybody - and it worked quite well, too (for a bit of cardboard).

If they stuck to what they built originally they would be dominating this segment right now.

> Kind of sad to see here on "hacker news" that 80% of the comments are low effort cheap shots

It's moreso that Google has used up all of its Goodwill a long time ago for that 80%, especially in the vr field.

The only people that will put investment into this are the 20% who don't remember every other time they've done the same thing.

Which basically means they worked on it, failed, and kept trying. Glasses too.

People tried Cardboard once and stopped using it.

Cardboard was great, and except a lack of software there were no problems about it in my opinion. I remember playing flight simulator on google earth and thinking how much potential this had. I have a meta quest 2 now and it is still not clear to me whether it is really that much better than cardboard.

It's possible for sure that Google will abandon this and I absolutely recommend anybody considering buying in do so only on the value they can see and realise immediately, not on any future promise.

But none of that takes away from the intellectually interesting part of this : what is new here, what possibilities does it open up? What implications does it have?

The main reason it's less likely this gets abandoned is because the spotlight in the the AI race is quickly moving to how much contextual information you can pour in about the user's ambient environment so that AI can actually do or say something useful for the user. That pretty much means glasses, and glasses mean you need a spatial computing OS to drive the underpinnings that the whole thing can operate on. Right now the technology for true AR glasses is still 2+ years out so the temporary placeholder for all that functionality is these larger headsets. But line of sight to all the pieces falling into place is there, so all the players are effectively in a long game where they are building up their ecosystems to be ready for the main game when it does arrive.

Cardboard was what got me interested in the whole area so it had it's value. But I think for about 90% of people it had negative value because it presented such a poor experience to the general user. To this day I ask people if they'd like to try my VR headset and they will say "no I tried that already, I know what it's like". Most of the time they mean cardboard and it bears almost no resemblance to the modern day experience. But 10 years later the impression still sticks.

There's VR and then there's VR, Cardboard was limited to 3DoF head tracking with a single button for input, which is not even remotely comparable to what we think of as VR today. Full 6DoF for head and hands has been table stakes for a long time, that's what you need to make something like Beat Saber or HL:Alyx work.

> I always felt this was such an outrageous burden to developers

but the point is that it's not a burden? You get it for free. Unless you mean having accommodate in your app the fact that someone else's might be "shading" it or similar.

I think it's amazing: you can have a real world light source coloring a virtual object which is then a reflective light source that bounces off to affect rendering of a second app. And you don't have to do any of it, the OS is rendering all of this. It's fully analogous to say, your OS supporting transparency on a 2d window frame such that if I'm looking at one window I can see the one behind it. But in 3d and incorporating real world pass through it is so much more complex.

If the Apple Vision Pro/Galaxy XR were viable than the MQ3 would be viable too, that is, the MQ3 does most of what the those do and it does more because it has game controllers and has decent applications and games, particularly fitness games. I don't think most people will pay a lot for "high definition" so getting people to pay almost 10x as much for something that's maybe 2x better is a hard sell.

I bought an MQ3 because I was curious about AVP and thought, "Hey I could get a six month head start on understanding XR application development" and came to enjoy the platform.

My complaint about the MQ3 as a software developer is that it has just 8 GB of RAM. With an AAA budget you can fit an AAA game into it, but it is a challenge to "share an experience with another VR user" based on photographic content and glTF models and whether you use Horizon Worlds or your own web site using

https://aframe.io/

and WebVR. It is straightforward to view that kind of content on a PCVR browser but to get it to work reliably on the Quest you have to be systematic about resource sizes.

The software innovation is real but it builds on the past. The MQ3 is basically an Android tablet you wear on your face. AVP is a Macbook Pro you wear on your face, etc. If you can use Unity Framework to make flat games you can use Unity Framework to make XR games.

In the 1990s I was a VRML enthusiast and got laughed at by all sorts of people who would say "So you're going to wheel down the aisles of the shopping center and put things in your cart?" Today we know that you can use a 2-d app store with VR controllers and it's great, it's great to use any web application which meets the WCAG AAA standard. You can just sideload phone and tablet applications into the MQ3 even system-y things like Tailscale and it frequently "just works".

I think Apple has thought through the "run 2d apps in a 3d space" a bit better than Meta did but the late rollout of controllers let MQ3 keep the lead in immersive apps. One of the titles that is packed in with the Galaxy XR is NFL PRO ERA which sent tingles up my spine on the MQ3 when I walked into a frickin' NFL stadium under the lights as the frickin' quarterback -- it was amazing.

That kind of hardware can deliver that kind of experience and Apple will have to catch up. Panographic photographic experiences can also be amazing in VR and Samsung is promising to deliver from Google and that's another selling point, but many MQ3 and AVP viewers now are watching and sharing panographic video on Youtube now.

I got my wife of all people to try a demo app on the MQ3 where cracks appear in your walls and your room gets invaded by aliens and you have to shoot them with the bop gun. She liked it.

But no way am I going to get her to sit through a cat simulator or Asgard's Wrath 2. She didn't like Beat Saber at all.

If it makes fake things feel real to the user, it’s worth it.

I think the lack of software that really took advantage of the possibilities and cared about the limitations — that wasn't simply a normal smartphone app with a bad UX because the display was now on your face — is the main reason Cardboard disappeared.

It's like: imagine if you just run the original DOOM in DOSBox on a phone and try to play it with the on-screen keyboard — that will obviously suck. Less obviously, even something as simple as going from a NES controller to an XBox controller can radically change experiences. You have to really consider what the right way is to use a system, and instead of doing that a lot of companies clearly go for existing zeitgeist in design language. (From memory as I heard it well before GenAI, real UX experts react to such UI designs in much the same way that artists react to Stable Diffusion).

Same goes for most VR stuff: There's some good games, but selling it as that means headsets have to be priced as consoles. That excludes the Android XR, and absolutely excludes the Apple Vision Pro.

> The main reason it's less likely this gets abandoned is because the spotlight in the AI race [...]

So we're assuming that it won't get killed in 5 years because it's nicely tied to the current bubble ?

I hear you, but I wouldn't put the blame on Cardboard.

The reason these people haven't tried anything else since Cardboard is because VR is still clumsy, expensive, of limited use and/or vomit inducing. I say that as my headset is still in active use after 5 years of owning it. In many respects I think nothing better than Cardboard has yet came out at this point for the ultra casual user.

Otherwise people were willing to give the Vision Pro a try because it was launched with much fanfare with a huge press focus, and I'd expect the Meta glasses to also have interest from people getting to try it.

These kind of big mainstream targeted events need to happen more often and stick in the news for people's perception of XR to move on.

You need to use their shader and lighting model and yeah, you don't have full control of the lighting at that point.

If you have existing assets its really not trivial at all to port them and get them looking right. Not impossible but not trivial.

Is this the LLM hallucinating or you? Google has a long history of abandoning things. Samsung isn’t doing themselves any favors by pricing this at $1799usd. This is a #metoo product. Nothing more. Where is the killer experience that Apple headset / Samsung headset / Meta headset provides? Being social? Watching a movie? c’mon! They have nothing. Now, if AI gets to the point where you can stream worlds, imagine movies, or paint the universe than maybe we might have something down the road but today - It’s about as useful as the Apple Vision which isn’t very useful.

PiMax and others at least know their lane. Simulation. These phone makers and social media companies aren’t vested in it other than to sell you ads to your eyeballs.

> The interesting thing here is the core of it, being Android XR and its deep AI integration

I'm not interested in the OS or "AI" at all. What I really want to know is if I can connect this to a regular PC/handheld via USB-C and use the headset as a primary/secondary display, and if so, is it good enough for gaming? The biggest issue with all these handheld gaming devices flooding the market is that the screen is tiny and most PC games aren't optimised for such a screen - but having a headset with a virtual big screen display like this could solve that problem. Unfortunately Samsung don't make this clear at all on the linked page.

Porn. Porn is the killer app. You’ll see.

Cardboard and Project Tango. They had the Quest - both low-end and high-end - before even Facebook did, let alone Apple, and ceded it for no reason. In fact, they canceled Tango FOR Cardboard, meaning that, instead of the world knowing Google for having the most advanced XR platform, they were known for having the cheapest one (albeit also the most accessible).

>Especially the part to do with compositing together multiple spatial apps living together in the same shared space and even interacting with each other (eg: one app can emit a lighting effect that will shade the other's rendering).

This is the killer app, but where do you see that capability?

At the risk of getting way too personal for HN, my wife and I couldn’t have sex for a while recently due to medical reasons. For my own sanity, I decided to look at what VR porn was available as I hadn’t checked it out since whenever the Quest 2 was released.

Woof. I think OnlyFans has taken away all of the good looking porn actresses (performers?) away from the large producers. Those large producers are the only ones that can make VR porn - or rather, I’m sure it’s possible for OnlyFans types to but probably not worth it.

Anyways, I’m not so sure about your statement. VR is not the right environment to “enjoy” a 4/10 on a good day.