Image Cropper
- No Sign-up
- 100% Free
- Zero Server Data
Drag a selection on the picture, or type the exact pixels, and cut out the part you want. Presets cover the ratios that actually get asked for: square for a profile, 16:9 for a slide, 9:16 for a phone screen. Your browser does the decoding and the cutting, so the photograph never leaves your machine, which matters more with a picture than with text because a photograph carries a date, a camera and often a location.
Drag inside the picture to move the selection, or drag a corner to resize it. With the picture focused, the arrow keys nudge by one pixel and Shift moves ten. The boxes below always show the exact crop.
Result
Choose an image to see the crop here.
Crop settings
That size is larger than the area you selected, so the picture will be scaled up. It will look softer. Scaling up spreads the same pixels over more of them and adds no detail.
JPEG has no transparency. The background will become white.
A circle needs transparency to keep its shape. Save as PNG or WebP, or the corners will be filled in white.
Which ratio for which place
| Ratio | Shape | Where it is used |
|---|---|---|
| 1:1 | Square | Profile pictures, avatars, product shots on most marketplaces |
| 16:9 | Wide | Slides, video thumbnails, most screens and TVs |
| 9:16 | Tall | Phone screens, stories, short-form video |
| 4:3 | Slightly wide | What most compact cameras and older screens produce |
| 3:2 | Wide | What most system cameras produce; standard photographic prints |
| 2:3 and 3:4 | Portrait | The same two, turned upright: posters, portraits, printed photos |
The ratio is a shape, not a size. Choosing 16:9 says nothing about how many pixels come out: that depends on how much of the picture you select, and the boxes under the image report it as you go. If a service asks for an exact pixel size as well, crop to the ratio here and then set a maximum width in the image converter, which will keep the shape you just chose.
Crop before you compress, not after
If a picture is going to be both cropped and made smaller, the order matters and only one way round is free.
- Crop first. You are throwing away pixels you had already decided not to keep, so nothing is lost, and everything after it has less work to do.
- Then resize and compress, once, to the size the picture is actually going to be displayed at.
Doing it the other way round means compressing pixels you are about to discard, and then re-encoding what survives a second time. Every pass through JPEG or WebP loses a little detail permanently, and two passes lose more than one. This tool always cuts from the file you chose rather than from its own previous output, so changing the selection or the quality re-encodes the original instead of compounding.
The part that is not a preview
The picture on screen may be scaled down to fit the page, or scaled up if it is small. That is display only. The selection is measured in the source image's own pixels. The four boxes show those numbers, and the crop is cut from the decoded image at full resolution.
So a selection reported as 1200 × 800 gives you a file that is 1200 × 800, whatever size the preview happened to be on your screen, and the pixels in it are the original ones rather than a preview that has been enlarged back up.
It also means a crop can be more precise than the pointer. Dragging gets you close; typing the numbers gets you exact, and both go through the same rules. A value typed into a box is clamped to the picture and held to the chosen ratio in exactly the way a dragged corner is.
What the tool will and will not do
It cuts a rectangle or a circle out, turns and mirrors the picture in quarter turns, straightens a crooked horizon by any angle up to 45 degrees, and re-encodes the result. Photographs from a phone are read with their orientation tag applied, so a picture your phone shows upright is cropped upright here too rather than on its side.
Straightening always costs you the edges. Turn a photograph two degrees and its corners swing outside the frame, so the tool keeps only the largest upright rectangle that is still entirely picture, which is why the frame gets smaller as the angle grows. The alternative is transparent wedges in the corners of the file you save, and you would find those afterwards.
It will not enlarge a small selection into a large file unless you ask for an output size bigger than the area you chose, and it says so when you do. The pixels for that do not exist, and inventing them softens the picture without adding anything.
A circular avatar usually does not need the circle crop, incidentally. Most services that show a round profile picture mask a square one themselves, so a 1:1 crop is enough. Reach for the circle when you need the round image itself (a logo to sit on a coloured background, or an avatar for somewhere that will not mask it for you), and save it as PNG or WebP, because JPEG has no transparency and will fill the corners with white.
Frequently asked questions
Is my photo uploaded anywhere?
No. Your browser decodes the file, draws the selected area to a canvas and encodes it again, all locally. The site’s security policy sets connect-src ‘none’, so the browser blocks network requests from these pages, meaning a photo could not be sent even if the code tried.
Can I crop to an exact pixel size?
Yes, and that is what the four boxes under the picture are for. They are not a read-out: type a number into any of them and the selection moves or resizes to match. It is the reliable way to hit an exact size, and it is also how the tool works without a mouse.
Does cropping reduce the quality?
The pixels you keep are the original pixels, taken from the decoded image at full resolution rather than from the preview on screen. The file is then encoded again, and that step is lossy for JPEG and WebP. Choose PNG if the picture must survive being cropped and saved repeatedly.
Why did the file get bigger?
A crop removes pixels, so it usually gets smaller, but not always. Re-encoding a heavily compressed JPEG at high quality, or saving a photograph as PNG, can produce a larger file from a smaller picture. The change is reported either way, including when it is negative.
What happens to the EXIF data?
It is dropped, as a side effect of re-encoding through a canvas: only the pixels survive, so the camera model, the timestamp and any GPS coordinates go. That is good for a photo you are about to post and a loss if you wanted the metadata kept.