Hacker News Reader

Comment by ss3000 | Hacker News Reader

Is VR not just a special case of AR?

IMO it's the other way around, AR is a special case of VR. To make an AR system, you need to basically do all the work involved in making a VR system, then somehow integrate it with a video feed of the outside world, then convincingly add digital content to that video feed in a way that complies with the real world's geometry, which you have to extrapolate from the video. Basically it's much easier to accurately place virtual objects in a virtual world than it is to put virtual objects in the real world. You know the geometry of the virtual world but can only make educated guesses about the geometry of the real world.

That’s why I’d think of VR as a special case of AR. As you said, VR is in many ways is “easier”. So I’d think of VR as a special case of AR where the environment is also computer generated.

In other words, moving from VR to AR would require a ton of extra work to integrate the real world. But moving from AR to VR should be much more straightforward, albeit in many ways equally challenging because now you have to worry about two worlds - the normal overlay of AR and the environment from VR.

Now that I’ve said that, maybe I should think of them as more like related disciplines instead of special cases of each other...

Depends if you're talking from the perspective of the hardware/OS developer, or the application software developer. For the system, AR is definitely harder. For the software developer, AR is "easier", if you're like most developers and are only treating an AR headset as a bad VR headset and just plunking 3D content down in some random place on the user's floor. You don't have to create the environment--the user supplies that. And you user's input modalities are so limited that you don't really have a lot of options on what sort of interactions you can design. For smartphone AR apps, you're basically just making a standard smartphone app that happens to have a 3D view in it. For headset AR apps, you're basically just making the "hunt-and-peck" version of what you could make on modern VR systems with much higher fidelity control systems.

When did AR become what you describe? I stopped at AR being overlaying data on top of the things you actually see - i.e. there's no "real world video feed", just semi-transparent displays.

I'm not sure what's going to happen first - technology to do proper AR like this, or fast enough, high-resolution (both spatial and color) enough cameras and displays being available to do the "AR via VR" option.

Not at all. They are completely different problems, both from technical and product perspectives.

VR is all about total immersion - sights, sounds, movements and everything else. The entire world needs to be contained in the headset, and what's happening outside it is irrelevant.

AR, on the other hand, is about interacting with your surroundings. Immersion is cut down, but the system now has to be intelligent about the outside world - recognizing objects, projecting at the right places etc.

I don't think Apple or Microsoft's (released) AR stuff can be used for VR. Apple's ARKit just runs on a phone screen (which obviously isn't AR) and HoloLens doesn't have the field of view to be VR. So they'll probably converge eventually but the products right now solve one or the other, but not both.

> which obviously isn't AR

Really? Why not?

Because it feeds the world through a camera onto a flat sheet of glass, instead of being a translucent surface onto which computer-generated images are projected.

My bad, meant to say "obviously isn't VR."

Reply on news.ycombinator.com