Why compress an image?
Oversized images are the most common cause of slow websites, bounced email attachments, and cloud drives that fill up faster than they should. The underlying reason is almost always the same: modern cameras and design tools produce images at resolutions intended for print or future cropping, and those images then get used, unmodified, in places that need a fraction of that data.
Compression fixes this by re-encoding the image at a lower quality and, often more importantly, a smaller resolution — while keeping it visually close enough to the original that the difference doesn't matter for its actual use.
Resolution matters more than quality settings
This is the single most useful thing to understand, and it's the step most people skip.
If you're displaying an image in a 800 px-wide column on a website, a 4000 px-wide source contains 25 times more pixel data than can ever be shown. No amount of clever quality tuning fixes that waste — you're compressing data that will be thrown away by the browser anyway. Reducing the dimensions to what's actually needed typically produces a far bigger saving than dropping the quality slider, and with no visible cost whatsoever.
Sensible targets: around 1920 px wide for full-width web imagery, 800–1200 px for in-article images, and 400 px or less for thumbnails. Double those figures if you're serving high-density retina displays and want maximum sharpness. The Medium preset in this tool caps width at 1920 px for exactly this reason.
Matching the format to the image
Different images compress well in different formats, and choosing badly wastes size or quality.
Photographs — organic detail, smooth gradients — compress efficiently with lossy encoding. WEBP or JPG both work well; WEBP is typically 25–35% smaller at the same perceived quality. Graphics with flat colour and sharp edges — logos, icons, diagrams, UI elements — behave differently: lossy compression produces visible halos around edges, so they need a higher quality setting or a lossless format. Screenshots with small text are the most demanding case; text legibility breaks down before photographic detail does, so compress these conservatively and check at 100% zoom.
What this tool does
Pick a compression level and drop in your JPG, PNG, or WEBP files. Each image is decoded and redrawn onto an off-screen canvas, then re-encoded as WEBP — or JPG, if your browser doesn't support WEBP encoding — at the chosen quality. Nothing is uploaded; the whole pipeline runs inside this browser tab, so client photography and unreleased design work stay on your machine.
Choosing a level
Low applies the gentlest compression with minimal visible change — the right choice for photography and portfolio work where image quality is the point. Medium balances size and quality and also caps resolution at 1920 px; this is the sensible default for web use and general sharing. High produces the smallest files with a visible quality cost, suited to thumbnails, previews, and situations governed by a hard upload limit.
A practical method: compress one representative image at Medium, view it at full size beside the original, and adjust one step if needed. Judging compression from a thumbnail is how blurry images end up shipped.
Compression is not reversible
Worth stating clearly: lossy compression permanently discards data. Re-saving a compressed image at a higher quality does not restore anything — it produces a larger file containing the same degraded image. Always keep your originals, and compress copies. This matters especially for anything you might need to re-crop, re-edit, or print later.
It also means you should avoid compressing the same image repeatedly across a workflow. Each pass compounds the loss. Compress once, at the end, at the size you actually need.