Celio

Encode any file into a video. Decode the video back into the original file, byte for byte.

Celio reads a file as raw binary data, renders it as a sequence of black-and-white pixel frames, and assembles those frames into a video. The decoder reverses the process — extracting frames, sampling pixels, and reconstructing the original bytes exactly. Error correction is built into the encoding so the data survives video compression.

Status: Work in progress.

The idea

A video is just a sequence of images. An image is just a grid of pixels. And a file is just a sequence of bits. Celio collapses these three facts into one pipeline: any file becomes a visual bitstream, and that bitstream can be played, stored, and reversed.

file  →  bytes  →  bits  →  pixels  →  frames  →  video
video →  frames →  pixels →  bits  →  bytes  →  file

Goals

• Encode any file type such as text, images, binaries and archives with no format-specific handling
• Perfect round-trip reconstruction, verified byte-for-byte against the original
• Resilience to standard video compression, not just lossless formats
• A clean command-line interface: celio encode <file> <video> and celio decode <video> <file>
• Written in C with minimal dependencies

Design principles

• Self-describing output. The video itself carries everything the decoder needs like the file size, layout, encoding parameters.
• Separation of concerns. Encoding, error correction, pixel rendering, and video I/O are independent modules that can be tested in isolation.
• Built in stages. Each stage of the pipeline is verified end-to-end before the next is added. Raw bit rendering first, then error correction, then video compression resilience.

Tech stack

• C — core encoding and decoding logic
• libpng — lossless frame output
• ffmpeg — video assembly and frame extraction (via subprocess pipe)
• Hamming codes — error correction for bit-level resilience