You cannot make a JPEG smaller without discarding something. What you can do is control what gets discarded — and most people throw away the wrong thing.
The order that actually matters
Almost every guide tells you to open a compressor and drag a slider. That is the last step, not the first. Do these in order and you will hit the same file size with a visibly better picture.
- Crop out what you do not need. Background is detail the encoder still has to describe. Cropping a portrait tight to the head and shoulders can halve the file size before you compress anything at all.
- Reduce the pixel dimensions to what the form actually uses. A 4000-pixel-wide camera photo displayed in a 300-pixel box is carrying thirteen times more information than anyone will ever see. Resizing the pixels removes data nobody was going to look at.
- Only now compress. With the image already at the right size, the compressor has far less work to do, so it can use a much higher quality setting to hit your target.
The difference is not subtle. A 2MB phone photo compressed straight to 50KB looks blocky. The same photo cropped, resized to 413 × 531, then compressed to 50KB looks clean.
Why a quality slider is a bad interface
JPEG quality is not a percentage of anything meaningful. Quality 60 on a flat studio portrait might produce 30KB; quality 60 on a busy outdoor photo might produce 300KB. Setting a quality number and hoping is why people end up compressing the same image four times.
Working backwards from the file size is the better model: tell the tool the ceiling, and let it find the highest quality that fits. That is what reduce image size in KB does — it encodes at several quality levels, measures each result, and narrows in on the best one that still clears your limit.
Things that quietly destroy quality
- Compressing twice. Each pass discards detail permanently and the losses compound. Always work from the original file, never from a copy you already shrank.
- Screenshotting instead of exporting. A screenshot captures your screen at screen resolution, throwing away most of the original image before you even start.
- Sending through WhatsApp. Normal image sends are recompressed hard, often down to 30–80KB. Ask for the file as a Document instead.
- Converting PNG screenshots of text to JPG. JPEG is built for photographs. Text and sharp lines go fuzzy. If the form accepts PNG for a document scan, keep it.
Greyscale for documents
If you are compressing a scanned certificate or a handwritten declaration and the form does not require colour, scanning in greyscale cuts the file size substantially with no readability cost — you are removing colour information that carries no meaning in a black-ink document.
When smaller is the wrong direction
Plenty of forms enforce a minimum as well as a maximum, and people who compress aggressively land underneath it. If you get a “file too small” error after shrinking, you do not need to start over — increase the size back up into the allowed band.
Frequently asked questions
How do I reduce image size in KB without losing quality?
Crop out unnecessary background, resize the pixel dimensions to what the form actually needs, and only then compress. Compressing a correctly-sized image loses far less than compressing an oversized one to the same file size.
Does reducing KB always reduce quality?
If the pixel dimensions stay the same, yes — JPEG is lossy and something has to go. But if you reduce dimensions you did not need, the visible quality at normal viewing size can be unchanged.
What is the smallest a photo can go before it looks bad?
For a cropped passport photograph, around 20KB is where visible blocking usually begins. Below that, reduce the pixel dimensions instead of compressing harder.
Is it better to resize or compress?
Do both, in that order. Resizing removes pixels you do not need; compressing degrades the pixels you keep. Resize first so compression has less damage to do.