How to compress images without uploading them

How in-browser compression works

When you drop an image into the Image Compressor, the browser decodes it into raw pixels and draws those pixels onto an invisible canvas element. The compressor can resize the canvas first if you set a maximum dimension, then asks the browser to encode the canvas into a new file. For JPG and WebP the browser's built-in encoder takes a quality value from 0 to 1, where lower numbers throw away more fine detail and give a smaller file. This is lossy compression: some information is discarded permanently, and good encoders discard the information your eyes are least likely to notice.

PNG is different because the format is lossless, so there is no quality dial in the encoder itself. The tool instead applies quantisation: it reduces the number of distinct colours in the image to a smaller palette, for example 192 colours instead of millions, and then stores the picture as an indexed PNG. Fewer colours means less data to store. Photographs with smooth gradients can show banding when the palette gets very small, while flat graphics, logos and screenshots often look identical at far fewer colours.

All of this happens in your tab. The file is read locally, the canvas and encoder run on your device, and the result is created as a download. No image data is sent to us.

Measured results from a real test

To avoid vague promises, we ran a test on a real photograph: a 4096 by 2160 pixel JPEG of 3.4 MB. We re-encoded it at several JPEG quality levels using the macOS ImageIO encoder, and we encoded a PNG version of the same picture with the UPNG library the compressor uses, at different palette sizes. Other encoders, including the one in your browser, will give somewhat different numbers, so treat these as indicative rather than exact. The source was already a compressed JPEG, which is typical of what people actually upload.

Resizing a very large photo usually saves far more than lowering quality, and a photograph belongs in JPG or WebP, not PNG.

What these numbers tell you

Three lessons stand out. First, the biggest single saving came from reducing the pixel dimensions. A 4096 pixel wide photo displayed in a web page or an email at 1000 pixels is carrying many times more pixels than anyone will see. Scaling it down to 1920 pixels wide and saving at quality 80 cut the file to roughly an eighth of the original, with no visible loss at normal viewing sizes on most screens.

Second, quality settings follow a curve of diminishing returns. Going from quality 90 to 80 saved a useful amount, and 80 to 60 saved more, but the drop below 40 is where soft edges and blocky artefacts start to become visible on detailed images. For photographs, a quality between 70 and 85 is a sensible range for sharing. For a thumbnail, you can go lower.

Third, photographs saved as PNG are very large. The same picture was more than twice the size of the JPEG at full colour. PNG earns its place for screenshots, line art, logos and anything with transparency or sharp text edges, not for camera photos.

JPG, PNG or WebP: which format to choose

The Image Compressor can keep the original format or convert to JPG, PNG or WebP, and the Image Converter handles format changes in bulk. Note that converting a PNG with transparency to JPG fills the transparent areas with white, because JPEG has no alpha channel.

• JPG is the safe default for photographs. It is supported everywhere and compresses photographic detail well, but it has no transparency and repeated re-saving gradually degrades quality.
• PNG is lossless and supports transparency, which makes it right for screenshots, logos and diagrams. It is a poor choice for photos unless you quantise the colours.
• WebP is generally smaller than JPG at similar visual quality and also supports transparency. Modern browsers and many apps support it, but a few older tools and some email or document workflows still expect JPG or PNG. If the file is going to someone you do not control, JPG is the least risky.

Step by step

1. Open the Image Compressor and add one or several JPG, PNG or WebP files.
2. Choose the output format. Leave it on the original unless you have a reason to convert.
3. Set a maximum dimension if the picture is far larger than you need. For web or email, 1600 to 2000 pixels on the long edge is usually plenty.
4. Move the quality slider and watch the before and after sizes. A value around 75 is a good starting point for photographs.
5. Zoom in on a detailed area of the preview to check for artefacts, then download the compressed files.

Privacy and EXIF data

Photos from phones and cameras often carry EXIF metadata: the camera model, the date and time, and in many cases GPS coordinates showing where the picture was taken. Posting a photo with its metadata intact can reveal your home or workplace without you realising. Platforms may strip some of it on upload, but you should not count on that.

Because the compressor redraws the image onto a canvas and encodes a new file, the new file normally does not carry over the original EXIF block. If your goal is specifically to clean metadata, use the Remove EXIF Metadata tool, which re-saves the picture on a clean canvas and is built for that purpose. Resizing and cropping also reduce what the image reveals. For sensitive parts of a screenshot, the Blur and Pixelate Image tool lets you obscure text or faces before sharing.

Limits to know about

Compression is lossy for JPG and WebP, so keep your originals. The tool does not produce AVIF files, and it does not recover detail that was never in the picture. Very large images, such as panoramas of tens of megapixels, may be slow or hit memory limits on older phones. Browser encoders differ slightly between Chrome, Firefox and Safari, so the same quality value can give a slightly different file size on each.

Frequently asked questions

Try these tools