Making Portrait mode yourself
summary
An iOS implementation that recreates portrait-mode background blur without using depth data by running on-device Core ML segmentation to produce an alpha mask, refining hair regions with alpha matting and inpainting, and applying selective Gaussian blur to background layers. The Swift code runs at interactive frame rates on recent iPhones, handles edge artifacts and semi-transparent hair, and includes quantitative examples and photo comparisons to the iPhone dual‑camera output.
AI-written summary · about
related talks
17mOptimizing Portrait Lighting at Capture-Time Using a 360 Camera as a Light Probe
22mEverything a Swift Dev Needs to Know About Machine Learning
22mDeeper into ARKit with CoreML and Turi Create
23mBring life to your iOS Application with 3D content using RealityKit
32mSelf-Portraits in JavaScript
24mSuper Resolution with CoreML
07mSuper Resolution with CoreML
25mAI-Assisted Swift in Gamifying an iOS App