What does an online image cropper do?
An image cropper keeps a rectangular part of a picture and throws the rest away. Unlike a resize, which changes how many pixels are used to describe the whole frame, a crop changes what the frame contains — the pixels outside the selection are not stored at all, so the file that comes back is genuinely smaller rather than a large image with the extra area hidden.
That difference is the reason cropping is the right first move for most images that need to shrink. A photograph from a phone might be 4000 pixels wide, and a profile picture needs 400; cropping to the head and shoulders before scaling removes the data that would otherwise be scaled down and re-encoded for nothing. This tool draws the crop on the original at full resolution and only then encodes the region you selected.
The selection can be made two ways, and both are useful. Dragging on the image is quick and visual, and the selection can be moved by dragging inside it. Typing the left, top, width and height is exact and repeatable — the approach to take when several images must share the same crop, or when the region has to start at a particular pixel. An aspect lock ties the width and height together so the result is a perfect square or a standard social size.
How to crop an image
- Choose an image. It is decoded in the browser and never uploaded, so large files are handled locally.
- Drag anywhere on the picture to draw a new selection, then drag inside the selection to move it.
- Or set the left, top, width and height as exact pixel values when the crop must be reproducible.
- Pick an aspect ratio — square for an avatar, 16:9 for a video thumbnail, 9:16 for a story — and the selection reshapes to match.
- Choose the output format and, for JPG or WebP, the quality, then press Crop and preview.
- Download the result, or adjust the selection and crop again; the original is never modified.
Image Crop limits and rules
| Item | Limit |
|---|---|
| Input formats | Anything the browser can decode, including PNG, JPG, WebP, GIF and AVIF |
| Per-file cap | 30 MB, because the whole image is held in memory while it is edited |
| Selection | Drawn by dragging, or typed as exact left, top, width and height values |
| Aspect ratios | Free, 1:1, 4:3, 3:2, 16:9 and 9:16, locked to the centre when applied |
| Output scale | 1×, 2× or 0.5× the selection size |
| Output formats | PNG, JPG and WebP, with a quality slider for the lossy two |
| Not covered | Rotating a crooked horizon, erasing objects and anything beyond the crop rectangle |
When to use it
- Cutting a square avatar out of a rectangular portrait photo.
- Trimming a screenshot down to the part worth sharing.
- Making a 16:9 thumbnail from a photo that was taken in another shape.
- Applying the same crop to a set of images so they line up in a grid.
- Removing a border or a watermark-free margin before compression.
Is it private?
The image is decoded, drawn and encoded entirely inside your browser using the canvas API. It is never uploaded, stored or logged, and no request leaves the page — the crop can be checked with the network tab open.
Frequently asked questions
Not by itself — the cropped pixels are the original pixels, moved to a new canvas. Quality is lost only if the output is re-encoded as a lossy format, and only slightly: a JPG saved at high quality loses a little each time, a PNG or WebP in lossless mode does not. So crop from the original file rather than from a copy that has already been edited, and prefer PNG where the image will be cropped repeatedly.
Because the discarded pixels are not written to the output. A crop covering a quarter of the frame produces roughly a quarter of the image data before compression, so even after re-encoding the file is much smaller. That is what separates a crop from a resize: a resize of the same picture would keep every pixel and change the scale instead, which is better when the subject fills the frame but worse when it does not.
Yes, in two ways. Typing the width and height gives you a selection of exactly that size, and the output scale option then multiplies it — so a 400 by 400 selection exported at 2× produces an 800 by 800 file. That is how you crop and upscale in one step when a service demands specific dimensions, though it is worth remembering that scaling up interpolates and cannot invent detail that was never in the original.
Because a crop never moves the subject — it only removes surroundings, and the square is placed where the selection is, not where the face is. Applying a ratio re-centres the selection on the same point, so if the person was off to one side the square may cut them. The fix is to drag the selection over the part that matters and let the ratio reshape it around that point, rather than applying the ratio first and moving afterwards.
PNG and WebP keep it; JPG cannot store transparency at all, so a JPG output is filled with white wherever the source was transparent. That matters when cropping a logo or an icon that has a transparent background — choose PNG, because a re-encode to JPG would put a white box behind artwork that was designed to sit on any colour, and the fill cannot be added back later.
No. The browser reads the file into memory, crops it there and produces a new file for download; the original on disk is untouched. Because nothing is uploaded, closing the tab discards everything, and the same source image can be cropped any number of different ways in one sitting without ever being overwritten.
Related tools
Written and maintained by Md. Nur Hossain Repon.