6 ms·
Very interesting insight from a former GPU developer at Magic Leap: https://imgur.com/a/hIXTQ#e6Vdo3B https://imgur.com/a/hIXTQ#e6Vdo3B He believes they shoul
by dmix 9y ago
Very interesting insight from a former GPU developer at Magic Leap:
https://imgur.com/a/hIXTQ#e6Vdo3B https://imgur.com/a/hIXTQ#e6Vdo3B
He believes they should have focused on making the headset and gone with integrating with a mobile device instead of an integrated SoC...That said there's usually a reason (low-level) engineers are engineers and not product people... but he makes some good points.
The one I most disagree with is "why develop Apps with ML when you could develop apps on Apple (iOS)"
Well one good reason is that a) game developers and VR/AR devs aren't necessarily developing on iOS already yet and b) it's an entirely new (potentially large) new market where you could become a top developer more easily. And they'd still have to build a significant part of their SDK on top of iOS or Android, no?
- simias 9y agoIt's also a very niche market and you end up completely captive in their ecosystem if you do this. It's reasonable to assume that there'll still be a market for iPhone and Android apps 5 years from now, I don't even know if Magic Leap One will still exist in 5 months or if they'll ever achieve commercial success. Yeah it'll probably be vastly easier to be the top ML developer but it won't do you much good if you end up with 12 users and a dead platform. IOs/Android development is reasonably mature now and they're backed by huge companies (and other smaller companies who make dedicated tools for that ecosystem). It would make sense to build on top of that instead of starting from scratch. Also, price-wise you'll never be able to compete with smartphones when it comes to bang-for-the-buck. The economies of scale are huge. These AR/VR need as much processing power as you can throw at them, it's very hard for a small team to come up with a competitive offering, especially for an embedded wearable solution. If they were making their own custom SoC with specialized hardware to enhance the AR experience it might make some sense, but as far as I understand they're only using off-the-shelf components. It does seem like a strange design choice.
- vanderZwan 9y ago> It's also a very niche market and you end up completely captive in their ecosystem if you do this. During my masters in Interaction Design I kind of "accidentally specialised" in gesture interfaces (the pointing kind, not the touchscreen kind). So obviously I have an old-school Kinect, and later got a Leap Motion[0]. This is exactly the problem with the latter. Your only real options are either doing research, or doing some kind of custom job like designing an AR/VR installation for a museum. [0] https://www.leapmotion.com/ https://www.leapmotion.com/
- fsiefken 9y agoYou can use the Leap Motion in AltspaceVR and Bigscreen already, I hope the Tabletop Simulator developers will support it natively with intuitive gestures one day.
- jpalomaki 9y agoWould it be really feasible to use either Bluetooth or WIFI as the connection between glasses and the CPU/GPU? Sounds like quite challenging setup if you want to provide a smooth user experience. There's a lot of data coming from the sensors on the headset and graphics data going back and much of this is pretty time critical. With iOS, you will likely also run into issues with background processing. On mobile phones, the battery life is always an issue if you plan to do CPU/GPU intensive stuff. Sounds much better approach to make the glasses smart and capable of working to some extent independently. Then you can have a higher level protocol between the glasses and phone.
- plasticchris 9y agoI would imagine the latency at the required (high) resolutions would make WiFi / BT unusable.
- zitterbewegung 9y agoA big barrier to VR is the fact that it doesn’t work out of the box and needs configuration. My VR headset is collecting dust since I don’t have time to reconfigure it. If you make a device that works as fast as a Nintendo switch (put it on head, turn on and it opens the game) you will have a Really compelling device .
- corndoge 9y agoEh? The setup process on Rift is a 5 minute operation, granted the Vive takes about twice as long. Don't see this as significant either way though, not many people drop $400-800 on a piece of kit and then don't bother to use it because they're too lazy to configure it.
- drcode 9y agoYou're ignoring having to mess around with your PC every time you want to use the Rift/Vive.
- Bjartr 9y agoWhat do you mean by "mess around"?
- drcode 9y agoWell, I don't own a Rift, but I assume you can't just walk up to your PC while it's in sleep mode or whatever and without touching it, put a Rift on your head and automatically continue playing a game you had been last playing two days ago (Something that they've promised the "Oculus Go" will be capable of)
- sbr464 9y agoExactly, the thought of having to move the sensors around for optimal performance (I can't leave the 2 rear sensors with cables on the floor out in the office), and recalibrate each time translates to the Oculus not being used.
- derangedHorse 9y agoThere's an AR kit for IOS which would make development super easy if you already know the basics behind IOS app development and are a game developer (which I imagine is a sizable population).
- criddell 9y ago> The one I most disagree with is "why develop Apps with ML when you could develop apps on Apple (iOS)" I could see that making sense if it means you can ship earlier. At this point, it's starting to feel like the AR/VR hype machine has flipped from breathless optimism to cynical skepticism. If ML ends up shipping some $2000-$3000 dorky looking headset people will laugh and the product will die.
- roymurdock 9y agoPresumably Magic Leap is developing hardware and software that will be highly optimized for AR/computer vision and will go far beyond the capabilities of an iPhone or Android device. If ML was to come out and say tomorrow they are relying on the iPhone to run their headsets and apps, their valuation would tank. It's built on the premise that ML has some type of secret sauce that will be much better and more immersive than what's out there on the market right now. If they were going to use a 3rd party dev ecosystem and build their SDK on top of an AR dev tool, I'd expect Unity would be the top choice. Apple didn't really have an AR-focused dev tool until they publicly released ARKit in the middle of last year, making that somewhat of a non-starter for a company that has been around as long as ML.