Bulk Image Compressor

How much smaller will my images get?

It depends almost entirely on what the file already is. A photograph straight off a phone or a camera is usually saved at close to maximum quality and will typically lose 60–80% of its size at 75% quality with no visible difference on a screen. A JPG that has already been through a compressor may barely move — and if re-encoding it would make it larger, this tool keeps the original file instead of handing you a bigger one.

PNGs are compressed too, with the PNG files setting rather than the quality slider. Smart reduces each image to a palette of up to 256 colours and recompresses it — typically 60–75% smaller for logos, product shots and screenshots, with no visible change. Lossless keeps every pixel and saves less, up to about 20%. If your PNGs are photographs, converting them to JPG, WebP or AVIF with the format control is often a bigger saving still, and avoids the banding a palette can cause in smooth gradients. There is more on both modes on the compress PNG page.

Quality or target size

Quality applies one setting to the whole batch. The file sizes that come out depend on each picture, which is the honest way to compress: the same quality looks the same across a folder of photographs.

Target size works the other way round, for upload forms and listings with a size limit. You give a size in KB, and each image is re-encoded at the highest quality that fits under it, found in up to seven attempts. It works for JPG, WebP and AVIF output only — a PNG has no quality setting to search, so with Keep original a PNG is refused and you choose JPG, WebP or AVIF instead. A target is not always reachable: a large or detailed photograph may still be over it at the lowest quality, and the row then says so and shows the closest size reached. Check the size each row reports rather than assuming the target was met. In both modes the width and height stay exactly as they were.

Compressing or resizing?

Compression makes a file smaller and leaves its width and height exactly as they were: each image is re-encoded — at the quality you choose, or at the highest quality that fits your target size — and when the format is kept and the result would not be any smaller, you get the original back. Resizing is a different operation — it changes the number of pixels in the image.

Which one you need depends on the problem. If the dimensions are right but the files are too heavy, compress. If the images are simply bigger than they will ever be shown — a 4000-pixel photo displayed 1200 pixels wide — most of that weight is pixels nobody sees, and the bulk image resizer is what removes them. For a folder of camera photos headed for a website, the answer is often both: resize first, then compress the result.

Why compress in the browser?

Compression is arithmetic on pixels your computer already has. Sending an image to a server to have that done, then sending it back, is slower, costs someone money, and means your files sit on a machine you do not control. Doing it locally is faster for a large batch, works offline once the page has loaded, and is the reason this tool has no sign-up and no usage limit.

Choosing between JPG, WebP, AVIF and PNG

• JPG — photographs. Universally supported, no transparency. The safe choice when the image is going somewhere you do not control.
• WebP — photographs and graphics. Typically 25–35% smaller than a JPG at the same visual quality, supports transparency, and is supported by every current browser.
• AVIF — the smallest of the four for photographs, often half the size of a JPG that looks the same. Every current browser shows it, but some older apps and upload forms do not accept it yet, and it takes longer to create.
• PNG — logos, screenshots, anything with sharp edges or transparency that has to stay pixel-exact.

Frequently asked questions

Are my images uploaded to a server?

No. Compression happens in your browser: images are read with the canvas API, and JPG, PNG and AVIF files are written by encoders compiled to WebAssembly that run in a background worker in the same tab. Your images never leave your computer, which is also why the tool works with no account and no limit on how many times you use it.

How many images can I compress at once?

Up to 200 images per batch. They are processed a few at a time rather than all at once, so a large batch will not exhaust your browser memory.

Which formats are supported?

JPG, PNG, WebP, BMP, GIF and AVIF. JPG, PNG, WebP and AVIF are written back as themselves; BMP and GIF are read and returned as JPG, PNG or WebP, because this tool has no encoder for those two. You can also convert everything to JPG, WebP or AVIF as you compress. Animated GIFs, animated WebPs and APNGs are refused rather than accepted, because a canvas only ever sees the first frame and the animation would be silently lost.

How are PNG files compressed?

Two ways, chosen under PNG files. Smart, the default, reduces the image to a palette of up to 256 colours and recompresses it, which typically takes 60–75% off logos, product shots and screenshots with no visible change; photographs with smooth gradients can show slight banding, so check those with Compare. Lossless keeps every pixel exactly and only redoes the compression, which saves up to about 20%. Either way, if the result is not smaller, the original PNG is returned unchanged.

What quality setting should I use?

For photographs on the web, 70–80% is usually indistinguishable from the original at a fraction of the size. Below 50% compression artefacts become visible in flat areas such as skies.

Why did one of my images not get smaller?

Some images are already compressed as far as they usefully go, and re-encoding them would make them larger. With the output format left on Keep original, that larger re-encode is thrown away and the original file is returned unchanged. Two cases can come back larger: BMP and GIF, which this tool does not write, so they are always re-encoded to JPG, PNG or WebP; and any image you convert by choosing JPG, WebP or AVIF as the output format, because the conversion you asked for is kept even when it grows. Each row shows the real change in size. In Target size mode the same rule keeps an original that the search could not beat, and a converted image that cannot get under its original size is reported as failed instead of being handed back bigger.

Related reading

Need to remove backgrounds rather than shrink files? ImageBulk’s bulk background remover processes whole batches with AI, and there is a guide to batch image processing if you are working out a workflow. Selling online? See Amazon’s photo requirements.