Image compressor — smaller files, no upload.

Shrinking a photo without uploading it

Every mainstream image compressor asks you to upload first. This one does not: the file is read by JavaScript in your tab, redrawn to a canvas at the size you choose, and re-encoded by your own browser.

1. Choose an image. JPG, PNG, WebP or a phone photo. The original stays on your disk untouched — the tool reads it, never writes to it.
2. Set the longest side. This is usually the biggest lever. A photo straight from a phone is often 4,000 pixels wide; very few places need more than 1,600. Halving the dimensions cuts the pixel count by three quarters.
3. Adjust quality. The slider trades file size against visible artefacts. Around 75–85% is the range where the saving is large and the difference is hard to see at normal viewing size.
4. Download. The result shows before, after, and the saving as a percentage. If the original was already small and well compressed, the tool says so instead of producing a larger file.

When this comes up

• Getting under an upload limit. Email attachments, job portals and forum uploads all have size caps, and a phone photo will exceed most of them.
• Making a page load faster. Images are the largest part of most page weight. A 4 MB hero image reduced to 200 KB is the single most effective performance change available to most sites.
• Sending photos that contain documents. A passport scan or a signed form is exactly the kind of file you should not be uploading to a random server to be compressed.
• Preparing images for a listing. Marketplaces and property sites impose their own limits and their own compression, and pre-shrinking avoids a double compression that looks worse than one.

A worked example: a 4.2 MB phone photo

A shot at 4032 × 3024 and 4.2 MB, compressed to 1920 px on the longest side at quality 80, lands near 350 KB — a 92% reduction that is invisible on any screen and acceptable in print up to about A5. The 4032-pixel original only earns its keep if you plan to crop hard or print A3.

Two settings do most of the work. Longest side is a hard resize: 1920 keeps a 16:9 photo at 1920 × 1080 and a portrait at 1080 × 1920, so the shape survives. Quality is the encoder's dial, and the curve is steep early and flat later — 90 to 80 saves about half the bytes, 80 to 70 saves a third as much again and starts to show on gradients and fine text.

Format is the third lever. WebP at matched visible quality is typically 25–35% smaller than JPEG and supported everywhere that matters now. JPEG remains the safer choice for anything that will be opened in an old desktop application.

Where compression goes wrong

• Compressing twice. JPEG is lossy, so every pass loses again. Always start from the original, never from a previously compressed copy.
• PNG to JPEG. Transparency has no JPEG equivalent and is flattened to black or white. Keep logos as PNG or WebP.
• Upscaling by accident. Setting the longest side above the original resolution adds no detail, only bytes.
• Metadata. EXIF is stripped, which removes GPS coordinates along with camera settings — usually what you want before posting a photo of your kitchen, occasionally what you did not.

Questions