arbvis

Visualize binary files as Hilbert curve plots. Each byte is mapped to a color and placed along a Hilbert curve, making structural patterns in the file visually apparent — null regions, ASCII text, compressed/encrypted data, and boundaries between file sections all produce recognizable visual signatures.

Quick start

arbvis huge_file.bin --tiles ./output
# then open output/index.html in a browser

Tiled viewer (recommended)

The primary way to use arbvis is the --tiles mode, which generates a Leaflet.js tile pyramid you can open in any browser:

arbvis file1.bin file2.bin --tiles ./output

This produces:

• output/tiles/{z}/{x}/{y}.png — XYZ tiles at multiple zoom levels
• output/index.html — ready-to-use Leaflet viewer

The tiled viewer offers significant advantages over the single-image mode:

• Full resolution at every scale — the highest zoom level preserves one pixel per byte with no downsampling; lower zoom levels are averaged so you can zoom out smoothly across arbitrarily large files
• Vector file boundaries — borders between files are drawn as crisp vector polylines that remain sharp at any zoom level, rather than being baked into raster pixels
• Accurate file labels — labels are positioned at the area-weighted centroid of each file's actual data region and rendered as HTML, so they're always legible regardless of zoom
• No size limit — works on files of any size; the tiled pyramid approach avoids the memory constraints of single-image rendering

Color scheme

Bytes are colored by range (based on Stairwell's approach):

Value	Color
0x00	Black
0x01–0x1F	Green (control characters)
0x20–0x7E	Blue (printable ASCII)
0x7F–0xFE	Red (high bytes)
0xFF	White

Single-image mode

For quick one-off inspection, arbvis can display a window or write a PNG directly:

# Display in a window
arbvis /bin/ls

# Write to a file
arbvis /bin/ls --output out.png

# Multiple files (boundaries marked in black)
arbvis file1.bin file2.bin --output out.png

# Pipe from stdin
cat /dev/urandom | head -c 65536 | arbvis

If no files are given, reads from stdin. For large files the image is subsampled to fit a 4096×4096 canvas, so detail is lost — use --tiles instead.

Building

Requires Rust (stable).

cargo build --release
./target/release/arbvis <file> --tiles ./output

Dependencies

• clap — CLI argument parsing
• fast_hilbert — Hilbert curve index-to-coordinate mapping
• image — Image construction and tile encoding
• show-image — Windowed display (single-image mode)