Rust Module Cuts Pixel-Art Copy Detection from 9 Days to 1 Second per Query

Developers at Pixagram, a pixel-art social network with an on-chain marketplace, built a Rust-based structural hashing system to detect near-duplicate artworks at scale. Every new artwork minted must be checked against up to 10 million existing pieces for similarities like recoloring, rescaling, or single-pixel edits — not just exact byte matches. The naive pixel-comparison approach cost roughly 80 milliseconds per image pair, translating to about nine days of CPU time per mint against the full corpus. The new method converts each image into a 1,024-byte canonical key — normalizing palette order, transparent pixels, and scale — enabling a comparison in just 115 nanoseconds, roughly 700,000 times faster. Unlike standard cryptographic or photographic perceptual hashes, the approach is purpose-built for pixel art's quantized, flat-color structure, neutralizing entire categories of copy attacks at hash time rather than query time.