Image Cropper

These two operations get confused constantly and they do different jobs. Cropping decides what is inside the frame. Resizing decides how many pixels that frame is made of.

When a form or a platform gives you a target like 400 by 400, it is stating both a shape and a size. The shape is the part that cropping handles. If you skip the crop and resize a 4:3 photo straight into a square, the image gets stretched and a face gets visibly wider - a mistake that is instantly recognisable and surprisingly common.

So: lock the aspect ratio to the target shape, crop to include what you want, then resize the result to the exact pixel count. Doing it the other way round means you resize, discover the shape is wrong, crop, and end up with fewer pixels than you needed.

The presets here set the aspect ratio for you. They are named for common social and web slots rather than exact rules, and platforms change their specs, so treat them as convenient starting shapes rather than authority.

This is underrated. File size depends on two things: how many pixels there are, and how much detail is in them. Cropping reduces both at once.

Removing the outer third of an image removes roughly half its pixels, and the parts you crop away are usually background - wall, sky, desk, floor - which is exactly the busy, high-frequency material a compressor spends most of its bytes on. A tightly cropped portrait routinely encodes to a fraction of the size of the same photo with a metre of room around it, at identical quality.

This is the fix people miss when they are fighting a file size limit. Turning the quality down makes the whole image worse. Cropping tighter usually makes it better, because most document and profile specifications want a tight crop anyway.

The crop is a re-encode, so the format you pick matters more than it looks.

PNG is lossless. Nothing degrades, transparency survives, and the file will be large if the content is photographic. It is the right choice for screenshots, logos, diagrams and anything with flat colour or sharp edges.

JPEG is lossy and the right default for photographs. The quality setting controls how much detail is discarded. Note that cropping a JPEG and saving as JPEG means a second round of lossy encoding on top of whatever the camera already did - the damage is usually invisible at high quality, but it is real and it accumulates if you keep re-editing the same file.

WebP does both lossy and lossless, keeps transparency, and generally reaches a given file size at better quality than JPEG. The catch is acceptance: plenty of upload forms and older software still reject a .webp outright, so it is a good choice for a site you control and a risky one for a form that checks extensions.

The SVG option is a wrapper, not a conversion. It embeds the cropped raster image inside an SVG document. That can be convenient when a system insists on an SVG file extension, but it does not make the image vector and it does not make it scale cleanly - the pixels inside are the same pixels.

Photographs of documents are a slightly different job. What you want is the page and nothing else: no desk, no fingers, no shadow of the phone.

Cropping tightly to the page edges before anything else helps three separate things. It makes the file smaller. It makes OCR noticeably more accurate, because the recogniser is no longer trying to find text regions in a photo of a table. And it makes the result look like a scan rather than a snapshot, which matters when a human is going to look at it.

The one thing cropping cannot fix is skew. If the page was photographed at an angle, the edges are not parallel and a rectangular crop will always leave some background in one corner or cut into the page in another. Photograph flat and from directly above, and the crop becomes trivial.