// Research

DarkCode.

A research project on offline data transfer: encoding large amounts of text into a single dense, multi-color image that can be carried from one screen to another camera — with no network, no accounts, and nothing sent anywhere.

140K+characters per image
99.96%decode accuracy at 140K characters
0bytes sent over the network

Why we are building it

Transferring a long text between devices normally means uploading it to someone’s server. DarkCode explores the alternative: a screen shows a dense multi-color image, a camera reads it, and the text is reconstructed locally. The data physically travels as light.

Current status

DarkCode is an active research project. The current decoder reaches 99.96% accuracy at 140K characters and near-lossless performance at 100K. The research version of the codec has been published as open source, and the technology builds on the same offline-transfer direction as DarkDrop.

Where it is headed

DarkCode is separate from our shipped products while we keep improving reliability on real-world cameras and screens. When it is ready, it will land where offline transfer matters most — inside Bardo.

Research project · in active development