Niantic is working on a large geospatial model that uses machine learning to connect and understand scenes worldwide. It’s pretty exciting stuff! Is anyone else interested?
Can someone explain this in simpler terms? It looks cool
Imagine you want to place a virtual object or text on something like a rock. Later, you’d want to see it in AR exactly where you put it. This model estimates where the camera is and where to pin that text using just one image. It makes AR possible without fancy gear, just a regular camera.
I was using Google Earth at work and thought about this. If we had the right cameras to capture 3D street scenes, it could really change how we use VR. Exploring cities like NYC in VR could be way more immersive than just gaming. It’d be a great tool for game development too.
3D maps could really help security. Imagine checking out a location before going there and knowing exactly where everything is. I also want facial recognition in my glasses to alert me about people with warrants. Picture a cop driving by, spotting a missing person, and stepping in to help. That’d be wild.
Microsoft was trying to make something like this years ago
I think this is more advanced than what Microsoft was working on back then 
Yeah, like Microsoft wasn’t working on this for a while either
I think it’ll be amazing when everyone has AR glasses and Niantic has the biggest platform for these experiences.
Is it gaussian splats?
The idea for this has been around for years. There was a TED talk about it. That company got bought by Microsoft and the project got dropped.
Complete with all the guessing algorithms that might glitch out.