Image Compressor

Image compression is a negotiation between three things: how big the file is, how sharp it looks, and how large it is displayed. Most people only touch the first two and forget that dimensions dominate everything. A 4000×3000 photograph compressed to perfection still weighs more than a 1200×900 version at mediocre quality — which is why this image compressor puts a size cap next to the quality slider instead of pretending quality is the only lever. Used in the right order, a simple image compressor clears most upload limits without visible damage.

The order to work in

When you compress images, the order of operations decides how much you can save:

1. Resize first. Match the image to its largest real display size. Retina screens want roughly double the CSS pixel width, so a 600-pixel column needs about 1200 pixels of image.
2. Pick the right format. Photographs go to JPEG or WebP lossy. Logos, charts and screenshots go to PNG or WebP lossless.
3. Then lower quality. Only after the dimensions and format are correct does the quality slider become the efficient lever.

Target-size mode: compress images to an exact size

Some situations have a hard ceiling rather than a preference: a 500 KB portal upload, a 100 KB hero image for a performance budget, a 2 MB email attachment. Type that number into the target field and the image compressor stops guessing on your behalf and starts measuring.

It works as a descending ladder rather than a clever search. The first attempt encodes at 90% and the result is weighed. If it is still over your number the next attempt runs at 80%, then 70, 60, 50 and finally 40. The moment an attempt lands under the target the ladder stops, so a photograph that only needed a nudge is never flattened by a setting it did not require. Target mode is what turns a quality slider into a real image compressor: it weighs every attempt instead of trusting a number.

Two consequences are worth knowing. First, a target overrides the quality slider completely — set both and the slider is ignored for that run. Second, 40% is the floor. If even a 40% encode is larger than your target you receive the 40% version rather than an error, because a slightly oversized file you can look at is more useful than a failure message. When that happens the fix is dimensional rather than tonal: set a maximum dimension of 1600 or 1200 pixels and run it again. Quality changes shave bytes; pixel counts move them in blocks.

What image compression actually removes

JPEG and WebP lossy compression works in the frequency domain. It keeps the broad shapes and tones that your eye latches onto and discards fine high-frequency detail that it mostly ignores. That is why smooth skies survive brutal compression while hair, foliage and text fall apart — those are almost entirely high-frequency information.

If your image is mostly text or thin lines, do not use lossy compression on it at all. Keep it PNG, or resize it instead. The image compressor will happily flatten a screenshot if you let it; the format choice is the guardrail that stops it. To compress images well is to discard what the eye ignores and keep what it latches onto — that is the whole discipline in one sentence.

Verifying the result

Always compare at 100% zoom, not fit-to-screen. Browsers downscale previews in a way that hides exactly the artefacts compression introduces. Open the original and the compressed image in two tabs, zoom both to actual size, then flick between them while watching one small thing — an eyelash, a brick wall, the edge of a caption. Damage that is invisible at fit-to-screen is obvious at 100%, and 100% is what a reader who taps to enlarge will see. That habit costs ten seconds, and it is the only reliable way to check what an image compressor actually did to the file.

The maximum-dimension cap

The maximum-dimension field caps the longest edge, whichever edge that happens to be. A 4000 × 3000 landscape photo capped at 1600 comes out 1600 × 1200; the same photo rotated to 3000 × 4000 comes out 1200 × 1600. You never have to work out which side is which and you never have to touch the ratio, because the other edge follows automatically. It is the strongest single control when you compress images for the web, because dimensions dominate everything.

It also only ever shrinks. A 900-pixel image with the cap set to 1600 is left at 900, so a mixed batch of screenshots and camera files can go through one setting without the small ones being blown up. That single field usually does more than any quality change: dropping 12 megapixels to 1.9 removes 84% of the pixels before the encoder has done anything at all.

What you get back

Each finished row shows the output dimensions and the exact before and after size, so you can judge file by file whether the trade was worth taking — an image compressor that reports its own numbers instead of promising them. The strip above the batch adds both figures up and prints the percentage saved across everything you dropped in. Results are renamed with a -compressed suffix, so beach.jpg arrives as beach-compressed.jpg and a PNG kept as PNG arrives as beach-compressed.png — your originals are never at risk of being overwritten.

Batch image compression, privately

Twenty files per batch, 25 MB each, and not one of them leaves the tab. That last clause is the point of a browser-based image compressor: the usual online route means uploading the very photographs you are trying to fit somewhere else, which inverts the purpose of the exercise. This one will compress images where they already are — on your machine — with no account, no queue and no watermark. Application portals, email attachments and CMS media libraries all impose caps; drop the batch, set a target, read the numbers, download the results. If the balance is not right yet, adjust and re-run — the originals are untouched either way.