How I Made Sharp 950x Faster (And Why It Matters After Bun Joined Anthropic)

Source: Dev.to

The Problem with Sharp

Sharp is a battle‑tested image library used in thousands of production apps, but it has several limitations that become painful in modern serverless and high‑concurrency environments:

• No native HEIC support out of the box
• Requires building libvips from source
• Custom Docker layers needed for AWS Lambda
• Complex CI/CD pipelines
• Slow metadata extraction (decodes the entire image)
• Heavy under concurrent load, not optimized for serverless

These issues caused slow processing, expensive servers, and poor user experience in an image‑heavy application.

Introducing bun-image-turbo

A Rust‑powered image processing library built for Bun and Node.js from day one.

• 100 % open source, MIT licensed, free forever
• Install with a single command:

npm install bun-image-turbo

• No custom builds or compilation required
• Native HEIC support included

Benchmarks

Metadata extraction

Sharp decodes the whole image to read dimensions:

10 MB image → decode 10 MB → extract width/height → done (3.4 ms)

bun-image-turbo reads only the header bytes:

10 MB image → read ~100 bytes header → extract width/height → done (0.004 ms)

Downscaling strategy

For large downscales, bun-image-turbo uses progressive halving with the optimal filter at each step:

• GitHub:
• npm:

What You Get

Getting Started

npm install bun-image-turbo

For questions, drop a comment in the repository issues. If you find this useful, consider starring the repo ⭐.