Image Compressor

Compress an image to 100 KB, 50 KB, or 20 KB

Most upload forms do not care how many pixels a photo has. They care how many kilobytes it weighs. A job portal that rejects anything over 100 KB, a school system with a 50 KB cap, and a visa form that asks for 20 KB are all asking the same question: can you make this file smaller without making the photo unusable?

The target-size field answers that directly. Type the limit, choose KB or MB, and the tool tests the image at its current dimensions. If the file will not fit even at a low quality, it steps the dimensions down by 20 percent and tests again. The result is the largest and sharpest version that still comes in under your limit, not a guess based on the slider.

How target-size mode works

JPEG and WebP are lossy formats: the encoder throws away detail to make the file smaller. The relationship between the quality setting and the final size is not linear, and it depends on the picture. A flat screenshot and a detailed photograph at the same quality setting can differ by several hundred kilobytes.

So the tool measures rather than estimates. It encodes the image at a high quality first. If that fits, it stops. If it does not, it tries a low quality to see whether the dimensions are viable at all, then binary-searches the quality range for the highest setting that still fits. Only when even the lowest quality is too large does it reduce the dimensions. That order keeps as much detail as possible.

When to resize instead of compress

If a form asks for a 2 × 2 inch photo at 300 dpi, the answer is 600 × 600 pixels, not a 4000-pixel camera photo compressed to 100 KB. Resizing first gives the encoder less work and produces a cleaner result. Use the image resizer for passport, ID, and social sizes, then come back here if the form also has a file-size cap.

The same applies to a scanned document. A 300 dpi scan of an A4 page is about 2480 × 3508 pixels. If the reader only needs to see the text, 150 dpi is usually enough, and the file can be a fraction of the size before compression even starts.

Which target should you choose?

There is no universal number, so use the form’s own limit. If it says 100 KB, aim for 100 KB. If it says 1 MB, there is no reason to compress to 100 KB and lose detail. When there is no stated limit, 100–200 KB is a sensible range for a web image, 50 KB for an email attachment, and 20 KB only for a small thumbnail or a very strict portal.

For a Philippine passport or ID photo, the dimensions matter first: 35 × 45 mm is 413 × 531 pixels at 300 dpi, and a 2 × 2 inch photo is 600 × 600 pixels. Start from those numbers, then compress only if the upload form also caps the file size. The passport photo size guide lists the dimensions for the Philippines, US, UK, and other countries.

Why the tool gets under the limit, not exactly on it

An encoder cannot hit an exact byte count on demand. It can only produce a file and tell you how large it is. The tool therefore returns the best result at or below your target. If it says 96 KB for a 100 KB limit, that is the highest quality it could find that still fits. A tool that promises exactly 100.0 KB is either padding the file or not measuring what it claims.

The number it reports is measured, not guessed

The saving shown next to your file is the difference between the size you brought in and the size that actually came out of the encoder. It is not an estimate derived from the quality slider, and the difference matters more than it sounds.

Those two things genuinely disagree. At a mid quality setting a flat graphic such as a logo, a screenshot or a diagram can come out larger than the original, because there was nothing wasteful to remove and the re-encode only added overhead. And a very high quality setting can undo a resize, producing a file bigger than the one you started with after asking for a smaller picture. A tool that estimated from the slider would promise a saving in both cases and be wrong. This one encodes the image and weighs the result, which costs a second pass and is the only way to be honest about it.

Start from the lowest quality that still looks right

Move the slider up and down and watch the measured figure, rather than aiming at a particular number on the slider. Find the lowest setting at which the picture still looks acceptable to you, and stop there. Below that point the saving is real and the picture is not worth it.

Scaling down is where most of the saving comes from

Cutting the longest edge to a maximum size removes actual pixels, and pixels are what a file is made of. It is a bigger lever than the quality setting and it is not a compromise at all, provided the smaller size is still big enough for where it will be seen. A photograph on a website rarely needs the resolution of one on a camera.

Order matters: reducing the dimensions first and then setting a modest quality gives a predictable result. A high quality setting after a large reduction can hand back some of the size you just saved.

It re-encodes, so metadata goes

The result is drawn to a fresh canvas and written out as a new file, so EXIF data including GPS coordinates is not carried over. For a photo you took at home or at work, that is often the point. If you need the original data preserved, keep the original file, because the compressed copy will not have it.

Pictures that do not compress

Images that are already heavily compressed, and images that are mostly one flat colour, have very little slack to remove. For a logo or a screenshot, a PNG is usually a better answer than a smaller JPG, and the WebP to JPG or PNG tool will tell you what the honest file size is before you commit.

Your original is untouched

Everything happens on a copy in the page. The file on your disk is not modified, so the compressed version is a second file rather than a replacement, and you can compare the two at full size before you use it.