• 0 Posts
  • 8 Comments
Joined 2 years ago
cake
Cake day: June 16th, 2023

help-circle





  • Yeah. You train it to recognize an object/image, and then you can respond to that however you want, programmatically. It’s really cool for things like tabletop games.

    For the how: Apple has APIs to including lightweight ML models for that sort of thing. You have to train a model, but for something like a card game it’s super easy. For physical objects it requires more prep just because you have to take photos instead of using existing artwork… but it’s still relatively straightforward.


  • I have a card game (a physical one, not virtual) and I want to “replace” the real cards with digital, animated, “living” ones. Ideally, I could apply this technique to other types of tabletop game, later… but cards are the current project.

    This works fine on iOS and even in the Vision Pro simulator, but on hardware, the image recognition is slow and unreliable, and it doesn’t track items through space in real-time. It’s laggy and “floaty”. Image recognition for unique, flat cards should be one of the simplest possible use-cases… and given how much more powerful the hardware is than a phone, and the fact that it works on the simulator, it sure seems like a software issue… but you can’t ship Apps with such severe problems, either.


  • As an “infinity monitor” you can use anywhere, it is really great.

    But it is not - as it was advertised - a “spatial computer”. I can’t even think of it as an AR device, because it is terrible at image recognition and tracking… something even an iPhone can do. I have no idea why that is the case, because the hardware is, theoretically, ridiculously powerful… but something is seriously, seriously wrong with the software right now… and it cripples the headset for what is supposed to be it’s one major use-case: spatial computing.

    P.S. In case it wasn’t obvious, I bought one to develop for, and as a developer I’m pretty angry at how poorly it performs at basic AR tasks.