They're not far enough apart. Stereo vision on humans is best inside 3-5 metres, usually, and by 10 metres out is not that great.
That's with pupils 65mm apart, give or take. Now scale that down to the horizontal distance between lenses on a phone. Coincidentally on mine they are about 13mm apart. You just multiply everything down linearly: my camera has the same experience at 60mm distance from an object as I do at 3m. It would be pretty useless past 2m, but also with those constraints you'd notice the quality drop off with range within a single object that was approaching 50cm depth.
Now that's OK itself, you could get some useful work done with that, but there are so many provisos to it that it would be very hard to sell as a feature. To top that off it's algorithmically painful. There's a reason the commercial 3D scanners don't (typically, only) use stereo pairs; there's almost always a better way to do it.
(oh, and to get out ahead of questions: Spatial Video capture on the iphone 15 apparently uses the LIDAR sensor for the depth map, not just the cameras)
That's with pupils 65mm apart, give or take. Now scale that down to the horizontal distance between lenses on a phone. Coincidentally on mine they are about 13mm apart. You just multiply everything down linearly: my camera has the same experience at 60mm distance from an object as I do at 3m. It would be pretty useless past 2m, but also with those constraints you'd notice the quality drop off with range within a single object that was approaching 50cm depth.
Now that's OK itself, you could get some useful work done with that, but there are so many provisos to it that it would be very hard to sell as a feature. To top that off it's algorithmically painful. There's a reason the commercial 3D scanners don't (typically, only) use stereo pairs; there's almost always a better way to do it.