15 December 2020 0 comments Node, JavaScript
I recently wrote a Google Firebase Cloud function that resizes images on-the-fly and after having published that I discovered that sharp is "better" than jimp. And by better I mean better performance.
To reach this conclusion I wrote a simple trick that loops over a bunch of .png and .jpg files I had lying around and compare how long it took each implementation to do that. Here are the results:
jimp▶ node index.js ~/Downloads Sum size before: 41.1 MB (27 files) ... Took: 28.278s Sum size after: 337 KB
sharp▶ node index.js ~/Downloads Sum size before: 41.1 MB (27 files) ... Took: 1.277s Sum size after: 200 KB
The files are in the region of 100-500KB, a couple that are 1-3MB, and 1 that is 18MB.
So basically: 28 seconds for jimp and 1.3 seconds for sharp
Don't ridicule me for my benchmarking code. These are quick hacks. Let's focus on the point.
sharpfunction f1(sourcePath, destination) {
return readFile(sourcePath).then((buffer) => {
console.log(sourcePath, "is", humanFileSize(buffer.length));
return sharp(sourcePath)
.rotate()
.resize(100)
.toBuffer()
.then((data) => {
const destPath = path.join(destination, path.basename(sourcePath));
return writeFile(destPath, data).then(() => {
return stat(destPath).then((s) => s.size);
});
});
});
}
jimpfunction f2(sourcePath, destination) {
return readFile(sourcePath).then((buffer) => {
console.log(sourcePath, "is", humanFileSize(buffer.length));
return Jimp.read(sourcePath).then((img) => {
const destPath = path.join(destination, path.basename(sourcePath));
img.resize(100, Jimp.AUTO);
return img.writeAsync(destPath).then(() => {
return stat(destPath).then((s) => s.size);
});
});
});
}
I test them like this:
console.time("Took");
const res = await Promise.all(files.map((file) => f1(file, destination)));
console.timeEnd("Took");
And just to be absolutely sure, I run them separately so the whole process is dedicated to one implementation.