AI repository intelligence

FR-AI / ANALYSIS

Why it is worth attention

It achieves at least 50x higher throughput than previous QR streaming prototypes by leveraging Rust-based RaptorQ fountain codes and fast_qr rendering compiled to WASM, making it production-ready for local offline transfers.

Who it is for

Developers needing fast, offline device-to-device transfer
IT professionals in air-gapped or secure environments
Users who frequently share files without internet or cables
Enthusiasts exploring innovative QR-based communication

Use cases

Transferring documents between two phones or a phone and laptop without any network
Sending text or files from a terminal (CLI) to a mobile device via camera
Quickly sharing files in environments where Bluetooth or Wi-Fi is unavailable or restricted
Demonstrating high-speed QR data streaming in education or tech demos

Strengths

Measured 50x+ throughput improvement over previous JavaScript-only QR streaming
Fully offline and PWA-ready – no upload server means full privacy
Rust-based RaptorQ and fast_qr compiled to WASM for high-performance encoding and rendering
Adjustable QR version, ECC level, FPS, and repair overhead to balance speed and reliability

Considerations

Transfer speed is highly dependent on device camera quality, lighting, and browser performance
Receiver must use the RaptorQR web app (or core library) – terminal CLI is sender-only
Large files (e.g., 6.5 MB) still take tens of seconds even at optimal settings

README quick start

RaptorQR

The world's fastest files and texts transfer between devices by displaying high-throughput animated QR codes and reading them with a camera.

Everything runs locally in the browser or terminal: no upload server, no Bluetooth, no cable.

RaptorQR started from an earlier open-source QR streaming prototype and has since become a substantial rewrite of the core transfer pipeline and user experience: FEC, QR rendering, worker scheduling, scanner integration, sender/receiver UI, CLI packaging, and the repo layout have all been rebuilt around a higher-throughput, production-ready architecture.

Live demo: https://qr.linkto.host/

Performance

RaptorQR uses the Rust cberner/raptorq implementation of RaptorQ (RFC 6330), compiled to WASM, as its primary fountain-code codec. This project also compiles erwanvivien/fast_qr to WASM for high-speed QR rendering, with a more feature-complete wrapper than the upstream WASM package, and uses ZXing WASM for scanning.

The result is a massive performance improvement over the original JavaScript-only transfer path. In measured tests, the new pipeline reaches at least 50x+ higher throughput in practical transfer scenarios.

Measured examples:

Scenario	Result
V20 QR, 4-code parallel playback, 30 FPS	up to 300 decoded QR symbols/s
V30 QR, 4-code parallel playback, 30 FPS	100+ decoded QR symbols/s
95.2 KB file transfer (V30-L x 4QR@30fps)	375 ms, about 254.0 KB/s
6.5 MB file transfer (V30-L x 4QR@30fps)	36 s, about 183.6 KB/s

The 95.2 KB and 6.5 MB file tests were measured on iPhone 16 / Safari as QR scanner as 'lab results'. Actual speed depends on device came

Description

Ultra-fast offline file and text transfer over animated QR codes with WASM fast_qr, ZXing and RaptorQ.

Related repositories

Similar projects matched by category, topics, and programming language.

Intuition-Lab

personal-model

Persome is an open-source tool that builds a private, local personal model on macOS from captured activity and provides evidence-linked context to MCP-compatible AI agents like Codex and Claude Code.

Python

1,239

TanStack

Featured

router

TanStack Router is a type-safe, data-driven React router with built-in caching, prefetching, and nested layouts, while TanStack Start extends it into a full-stack SSR framework.

Web DevelopmentFrontend Frameworks

14,861

vercel-labs