Mobiconf 2018: Tobias Due Munk, Developmunk
summary
Shows how iOS delivers live depth as AVDepthData backed by a CVPixelBuffer of half float (Float16) disparity, and how to read or write values without a full 32-bit convert. Describes syncing depth with color via AVCaptureDataOutputSynchronizer, ARKit face anchors with 50+ blendShapes and eye gaze, and practical limits: VGA-ish front resolution, quarter-resolution back for live, ~15 cm close limit, poor back-camera low-light performance, and filtering that fills holes but hides invalid pixels. Provides Core Image recipes (CIAreaMinMax + CIColorMatrix) for GPU normalization, CI depth blur for portrait effects, demos that control sound with hand distance, and references Halide and an app for exporting/editing depth maps.
AI-written summary · about
chapters (5)
related talks
43mInto the Deep – or what would the 80s have done with depth sensing technology?
15mDepthLab: Real-time 3D Interaction with Depth Maps for Mobile Augmented Reality
15mDepthLab: Real-time 3D Interaction with Depth Maps for Mobile Augmented Reality
54mMe, Myself and eyeOS
12mIncrease AR realism and precision with Depth API
18mDeep-dive on ARCore | Fundamentals of designing AR apps with machine learning models
31mARCore
40mLearn By Doing. Understanding spatial computing one small app at a time