@carboneio/image-size

Fast, lightweight NodeJS package to get dimensions of any image file or buffer.

Forked by Carbone

After a thorough security audit (pentest) of our dependencies, we decided to fork this project (image-size on GitHub and image-size on Codeberg) to fix several vulnerabilities and, in the process, optimize the library. For a detailed list of changes, please see the full changelog.

We started by ensuring the test suite had 100% code coverage before making any changes. We also compared our results and performance with the tests from sharp, a widely used, more general-purpose library.

We would like to warmly thank the original developer of this library for their excellent work. We are maintaining this fork only because we needed to apply some urgent fixes and improvements; as soon as the original project incorporates these corrections, we will be very happy to remove this fork and return to the main repository.

Key Features

• Zero dependencies
• Supports all major image formats
• Works with both files and buffers
• Minimal memory footprint - reads only image headers
• ESM and CommonJS support
• TypeScript types included
• Tested via GitHub Actions CI on Node.js 18, 20, 22, 24, and 26

Supported formats

• BMP
• CUR
• DDS
• GIF
• HEIC (HEIF, AVCI, AVIF)
• ICNS
• ICO
• J2C
• JPEG-2000 (JP2)
• JPEG
• JPEG-XL
• KTX (1 and 2)
• PNG
• PNM (PAM, PBM, PFM, PGM, PPM)
• PSD
• SVG
• TGA
• TIFF
• WebP

Installation

npm install @carboneio/image-size
# or
yarn add @carboneio/image-size
# or
pnpm add @carboneio/image-size

Usage

Passing in a Buffer/Uint8Array

Best for streams, network requests, or when you already have the image data in memory.

import { imageSize } from '@carboneio/image-size'
// or
const { imageSize } = require('@carboneio/image-size')

const dimensions = imageSize(buffer)
console.log(dimensions.width, dimensions.height)

Reading from a file

Best for local files. Returns a promise.

import { imageSizeFromFile } from '@carboneio/image-size/fromFile'
// or
const { imageSizeFromFile } = require('@carboneio/image-size/fromFile')

const dimensions = await imageSizeFromFile('photos/image.jpg')
console.log(dimensions.width, dimensions.height)

Note: Reading from files has a default concurrency limit of 100 To change this limit, you can call the setConcurrency function like this:

import { setConcurrency } from '@carboneio/image-size/fromFile'
// or
const { setConcurrency } = require('@carboneio/image-size/fromFile')
setConcurrency(123456)

Reading from a file Syncronously (not recommended) ⚠️

v1.x of this library had a sync API, that internally used sync file reads.

This isn't recommended because this blocks the node.js main thread, which reduces the performance, and prevents this library from being used concurrently.

However if you still need to use this package syncronously, you can read the file syncronously into a buffer, and then pass the buffer to this library.

import { readFileSync } from 'node:fs'
import { imageSize } from '@carboneio/image-size'

const buffer = readFileSync('photos/image.jpg')
const dimensions = imageSize(buffer)
console.log(dimensions.width, dimensions.height)

3. Command Line

Useful for quick checks.

npx @carboneio/image-size image1.jpg image2.png

Multi-size

If the target file/buffer is an HEIF, an ICO, or a CUR file, the width and height will be the ones of the largest image in the set.

An additional images array is available and returns the dimensions of all the available images

import { imageSizeFromFile } from '@carboneio/image-size/fromFile'
// or
const { imageSizeFromFile } = require('@carboneio/image-size/fromFile')

const { images } = await imageSizeFromFile('images/multi-size.ico')
for (const dimensions of images) {
  console.log(dimensions.width, dimensions.height)
}

Using a URL

import url from 'node:url'
import http from 'node:http'
import { imageSize } from '@carboneio/image-size'

const imgUrl = 'http://my-amazing-website.com/image.jpeg'
const options = url.parse(imgUrl)

http.get(options, function (response) {
  const chunks = []
  response
    .on('data', function (chunk) {
      chunks.push(chunk)
    })
    .on('end', function () {
      const buffer = Buffer.concat(chunks)
      console.log(imageSize(buffer))
    })
})

Disabling certain image types

import { disableTypes } from '@carboneio/image-size'
// or
const { disableTypes } = require('@carboneio/image-size')

disableTypes(['tiff', 'ico'])

JPEG image orientation

If the orientation is present in the JPEG EXIF metadata, it will be returned by the function. The orientation value is a number between 1 and 8 representing a type of orientation.

import { imageSizeFromFile } from '@carboneio/image-size/fromFile'
// or
const { imageSizeFromFile } = require('@carboneio/image-size/fromFile')

const { width, height, orientation } = await imageSizeFromFile('images/photo.jpeg')
console.log(width, height, orientation)

Performance

image-size reads a header; sharp decodes an image. That makes them hard to compare on anything but this one question: how long does it take to learn an image's dimensions? They give the same answer: across sharp's own 501 test fixtures, the two libraries agree on all 496 that both can read.

Measured against sharp 0.35.4 / libvips 8.18.6 on Node 24, macOS arm64.

From a buffer already in memory, parsing cost on its own:

From a file path, what an application actually pays, I/O included:

The second table is the one to plan with. Reading the file dominates the cost, so the two orders of magnitude on a buffer shrink to between 1.3x and 9.2x once I/O is counted. image-size also has no native binary to install and no start-up cost, where sharp pays about 50 ms on its first call.

libvips' cache is disabled for these numbers, since the benchmark re-reads the same fixture in a loop and would otherwise measure memoisation rather than work. Files are warm in the OS page cache on both sides. Run npm run bench for throughput across all thirty formats.

Limitations

1. Partial File Reading

   • Only reads image headers, not full files
   • Some corrupted images might still report dimensions

2. SVG Limitations

   • Only supports pixel dimensions and viewBox
   • Percentage values not supported

3. File Access

   • Reading from files has a default concurrency limit of 100
   • Can be adjusted using setConcurrency()

4. Buffer Requirements

   • Some formats (like TIFF) require the full header in buffer
   • Streaming partial buffers may not work for all formats

Development

npm ci          # install dependencies from package-lock.json
npm test        # unit tests, with a 100% coverage threshold
npm run bench   # detection throughput, per format, on small and large files

npm run bench accepts --filter=<substring> to benchmark a single format, e.g. npm run bench -- --filter=webp.

License

MIT

Credits

not a direct port, but an attempt to have something like dabble's imagesize as a node module.

Contributors