Add images
Private by design. Compression happens entirely in your browser using
canvas and Web Workers. No image is ever uploaded.
Updated 2026-11-07
Shrink JPG, PNG, and WebP images. 100% in your browser — files never leave your device.
100% in your browser. Nothing uploaded.
Private by design. Compression happens entirely in your browser using
canvas and Web Workers. No image is ever uploaded.
The browser decodes the image, optionally changes its dimensions, and encodes the result with the selected format and quality. Compression does not guarantee a smaller file: a tiny or already-optimised image can grow because a new container adds overhead. Judge the actual output bytes and dimensions shown by the tool.
Photographs usually tolerate a lossy format after a visual check; screenshots with text, logos, and alpha often need a lossless or transparency-capable output. “Quality 80” is not comparable across encoders or source images. Preserve the original until the output opens correctly, has the expected dimensions, and is smaller enough to justify the re-encode. If it is not, the correct result is to keep the original.
When comparing encoders, hold the source pixels constant and record input MIME type, input bytes, output MIME type, output bytes, width, height, and quality. A byte reduction without the dimensions is incomplete evidence. For a site asset, check the actual rendered page after compression because browser decode, CSS sizing, and responsive variants affect the result visitors experience.
“Why did the file grow?” Usually the new encoding overhead or a quality setting. “Is WebP always smaller?” No; content and encoder matter. “Can I recover original quality?” No lossy encode can restore discarded information. Keep an untouched original and treat the compressed file as a derived delivery asset.
For privacy-sensitive images, browser-local processing limits upload exposure, but downloaded files can still contain the pixels and metadata produced by the browser. Check the final file before sharing it, and do not assume a smaller byte count removes identifying content.
For delivery, pair the compressed asset with an appropriate alt description and a responsive width strategy. Compression improves transfer cost; it does not replace image dimensions, content description, or accessibility review. Compare the resulting page load with real device conditions where performance matters. Keep a naming convention that distinguishes source masters from derived web files.
Optimize your images for the web without compromising your privacy. Our free image compressor allows you to shrink JPG, PNG, and WebP files directly in your browser, meaning your images are never uploaded to any server. Reduce file sizes by up to 80% with our bulk processing tool, and preview the results in real-time before downloading. Perfect for improving website speed and Core Web Vitals.
For automated bulk workflows, image CDNs and on-the-fly resizing, try Cloudinary.
Try Cloudinary →Learn how pixels, aspect ratios, formats, and quality settings work in our guide to shrinking photos for the web.
No. Every file you drop is re-encoded in a Web Worker inside this tab using the vendored browser-image-compression library. The compressed result is generated locally; nothing — original or compressed — is ever sent to a server.
JPG, PNG, and WebP for input. By default the compressor keeps each file in its original format; toggle Convert to WebP if you want smaller, modern output (WebP is supported by every current browser).
It depends on the source. JPGs typically shrink 30–70% at quality 70–85 with no visible loss; PNGs with photographic content can drop dramatically when converted to WebP. The before/after panel shows the exact saving per file.
Yes. Drop a folder's worth of images and they queue up. When more than one file is queued, the Download button switches to a bulk ZIP download (built locally with jszip); a single file goes straight to a direct download.
Yes — re-encoding a photo through this tool drops the embedded EXIF metadata, including GPS coordinates, camera model, and timestamps. That is usually what you want when posting an image publicly.
Yes, after the first visit. The compressor and the vendored libraries are cached by your browser, so you can drop images on a plane or a closed network and the queue still runs.
Drop JPG, PNG, or WebP files into the drop-zone. The browser re-encodes each file with the quality you choose, using the browser-image-compression library in a Web Worker. Nothing is uploaded — the original and the compressed copy both live only in memory on your device. Close the tab and they are gone.