What “without losing quality” really means

Every JPG has already lost something; the question is whether you can see it. At a given file size, the bytes can go on more pixels or on more detail in each pixel, and the trick is to spend them where the eye looks. A phone photo straight from the camera is often 4000 pixels wide and 3–5 MB. Squeezing all of those pixels into 200 KB forces the encoder to throw away detail in every one of them, and the result has blotchy skies and smeared edges.

Shown on a web page or in an application form, though, that photo is displayed a few hundred pixels wide. Scale it down to 1200–1600 pixels first and there are roughly an eighth as many pixels to pay for, so each one gets enough bytes to stay clean. Downscaling is the step that makes everything after it painless, and on a screen nobody can tell.

The order that works

1. Resize to the size it will be shown at. For a form or a website, 1200–1600 pixels on the long side is plenty; for a profile photo, 400–800. Resize Image does it, or use the Limit dimensions option in Compress Image.

2. Save as JPG. For photos, JPG is the format every upload form accepts. PNG is lossless and keeps a photo several times larger, so a photo saved as PNG almost never fits a small limit. WebP and AVIF are smaller still, but many forms reject them: use them on your own website, not for portals. The best image format for the web explains when each one wins.

3. Let the compressor search. In Compress Image, tap 200 KB under the size box, or type the limit the form gives you. Instead of making you guess a quality number, it tries quality levels until it finds the highest one under the limit, and reduces the dimensions only if quality alone can’t get there.

4. Look before you send. Zoom in on edges, text and faces. If you can see blocks, the limit is too tight for that many pixels: reduce the dimensions a step and compress again from the original, not from the compressed copy.

Rough sizes to expect

These are typical figures for an ordinary photo saved as a JPG at a quality where the loss is hard to see. Busy scenes such as foliage, gravel or crowds come out larger; plain backgrounds and studio portraits come out smaller.

Under 500 KB: easy at 2000–2500 pixels on the long side. Under 200 KB: comfortable at around 1200–1600 pixels. Under 100 KB: around 1000 pixels. Under 50 KB: around 600–800 pixels, enough for an ID-style portrait. Under 20 KB: a small portrait or a signature a few hundred pixels across.

If your limit and your required dimensions don’t fit these ranges, quality has to give. A form that demands a 2000-pixel image under 50 KB will get a visibly compressed one, whatever tool you use.

Exam forms, visas and signatures

Government and exam portals usually give both a size band and dimensions: a photo of 20–200 KB at a set number of pixels, say, and a signature of 10–20 KB. A file that is too small is rejected just like one that is too big, so aim inside the band rather than as far under it as you can.

Check what your portal asks for first. Photo sizes lists the current rules for common passports, visas and exams, each with its official source, and the photo size checker tells you whether a file already passes. For signatures, compressing a signature to 20 KB covers the scanning and cropping that make it fit.

Mistakes that make it worse

Compressing a compressed copy. Each round adds new damage on top of the old. Always start again from the original.

Screenshots of photos. A screenshot is a PNG of your screen, often larger than the original photo and at the wrong size. Use the photo file itself.

Sending HEIC from an iPhone. Many forms won’t take it at all. Compress Image saves an iPhone photo as a JPG when Save as is left on Same format, or you can convert it first with HEIC to JPG.

Forgetting what is inside. Photos carry metadata, sometimes including where they were taken. It costs only a few kilobytes, but for a public post you may want it gone: Remove Metadata strips it, and what your photos leak explains what is in there.