Hacker News Reader

Curious about a couple points:

- How does the 3D scanning work? Presumably photogrammetry? What kind of accuracy are we talking about here in terms of typical measurement tolerances?

- The frames are actually custom-fit? Are they machined or 3D printed or do you just have access to many permutationsof a given frame?

[edit, formatting]

View on HN

Great question: -for older devices without a depth sensor, we do a combination of pure computer vision (detect landmarks in 2D, then solve for 3D, then apply a morphable model (3DMM). After an order is placed, we'll do photogrammetry to further refine the model. -for newer devices with the TrueDepth sensor (used for FaceID), we build a mesh directly from that. -accuracy is sub-millimeter. -frames are 100% custom. For each frame, for each customer, we adjust over 20 different dimensions for a perfect fit. Down to the unique contours of your nose. It's all 5-axis CNC machined (for plastic) and laser-cut (for stainless steel) (also sorry if I overstepped earlier. I'm just trying to be helpful. Long time reader, but first time commenter.)

Doesn't Apple limit the resolution of depth sensor data for 3rd-party apps? Such scans are much less accurate than FaceID or a 3D camera connected to a laptop or mobile device.

Newer iphones now have an IR projecting depth camera built in

Reply on news.ycombinator.com