# Apple Frameworks for TouchDesigner

A collection of native TouchDesigner operators that expose Apple's on-device machine learning and media frameworks, such as Vision, Core ML and RealityKit, directly inside TouchDesigner.

This project packages Apple's on-device frameworks, including Vision, Core ML, Core Image, VideoToolbox and MetalFX, SpeechAnalyzer, Sound Analysis, Natural Language, FoundationModels, ScreenCaptureKit and RealityKit, as native custom operators for TouchDesigner. It runs only on macOS and Apple Silicon, needs no external runtime, Python environment or cloud API, and inference runs asynchronously on worker threads so it never blocks TouchDesigner's cook cycle.

Where an operator replaces a stock TouchDesigner operator that otherwise needs Windows and an NVIDIA GPU, its channel and texture formats match the original as closely as possible. Each plugin ships with its own documentation covering parameters, output formats and measured performance, and a demo project provides a minimal usage example for every operator, covering tasks such as pose, hand and face tracking, monocular depth, OCR, subject cutout and an on-device language model.

Website: https://github.com/sygnalinc/Apple-Frameworks-for-TouchDesigner