Crop a photo by dragging a box or by typing exact pixels. Lock the shape to 1:1, 4:5, 16:9, or any ratio you type, cut a circle with transparent corners for a profile picture, resize the result if you need to, and download PNG, JPEG, or WebP.
Or paste one with Ctrl+V (⌘V on a Mac). PNG, JPEG, WebP, GIF, AVIF, BMP, and SVG are recognised from their own leading bytes, so a renamed screenshot still loads.
25 MB maximum · 40 MP source · 8,192 px output · never uploaded
Level a crooked horizon or turn a sideways scan before you crop, so the box frames the picture the right way up.
When the whole picture should stay in frame and only its pixel dimensions need to change, resize it instead of cropping.
Fit one picture to each platform's sizes in a single pass, with safe-area guides and a ZIP of every export.
Cropping does not change a single pixel inside the box; it decides which pixels to keep. This page stores the crop as four whole numbers in the picture's own coordinates: how far from the left, how far from the top, how wide, and how tall. The stage you drag on is scaled to fit the panel, so every pointer movement is converted from screen pixels into source pixels before it touches the crop, and every rectangle is rounded to whole pixels and kept inside the picture before it is stored. The fields, the overlay, the result preview, and the export all read those same four numbers, which is why typing 400 into the width field and dragging a handle until the readout says 400 produce exactly the same file. At the crop's own size those pixels are drawn one for one onto a new canvas with no resampling, so an ordinary 8-bit sRGB picture comes out unchanged as a PNG, while a wide-gamut or 16-bit source is converted to 8-bit sRGB on the way. Beyond that, only an optional resize or a lossy format changes them.
Drop a file on the panel, choose one, or paste a screenshot with Ctrl+V (⌘V on a Mac). The format is read from the file's own leading bytes, the browser applies the photo's EXIF orientation as it decodes, and everything happens in this tab. Nothing is uploaded.
Drag the box to move it, drag any of its eight handles to resize it, or press outside it and drag to draw a new one. Pick a ratio such as 1:1, 4:5, or 16:9 to lock the shape, switch on the circle for a profile picture, or type the exact X, Y, width, and height in pixels. With the box focused, the arrow keys move it one pixel at a time.
Keep the crop at its own size, or fit it into a box such as 400 × 400 for an avatar. The preview shows the final dimensions and an estimated file size. Choose PNG, JPEG, or WebP and download a file named after yours, such as photo-cropped.png.
X, Y, width, and height are fields in the picture's own pixels, kept in step with the box while you drag. Type a value that does not fit and it is pulled back inside the picture with a note saying what changed, rather than silently becoming a different number. A width that fits the picture but not at the current position moves the box instead of shrinking it.
Free, 1:1, 4:5, 3:2, 4:3, 16:9, 9:16, and 2:3 are one click each, a custom ratio accepts decimals such as 1.91:1, and one button swaps portrait for landscape. The height is always derived from the width, so a locked box stays within half a pixel of its ratio however far you drag, and changing the ratio keeps the box's size and centre wherever the picture allows.
Circle mode locks the box to a square, shows the circular mask on the stage and in the preview, and exports the four corners as true transparency. Those corners are about 21.5% of the square, which is why JPEG is switched off in this mode: it has no alpha channel, so the corners would turn into solid colour.
At the crop's own size the pixels are copied across one for one with no resampling, so for an ordinary 8-bit sRGB picture a PNG export keeps the selected pixels unchanged. The browser canvas does convert a wide-gamut photo (Display P3 or Adobe RGB) to sRGB and save a 16-bit PNG at 8 bits per channel, and faint semi-transparent pixels can shift slightly. Resizing is a separate, optional step that keeps the crop's shape, never stretches it, and warns you when it would enlarge the picture, because enlarging adds no detail.
The box is driven by pointer events, so it drags the same way with a mouse, a finger, or a stylus. From the keyboard, the arrows move it 1 px, Shift moves it 10, and Alt (Option on a Mac) with the arrows grows or shrinks it. Undo steps back through recent changes, and a run of nudges counts as one step.
Nothing leaves the tab. The export is a fresh canvas encode, so EXIF metadata such as GPS location and camera model is not carried over. The caps are 25 MB per file, 40 megapixels of source, and 8,192 px on the longest output side, and anything past them is refused with the number that broke it.
Load the picture by dropping it on the panel, choosing it, or pasting it. A box appears over the image with everything outside it dimmed. Drag the box to move it, drag its corner or edge handles to resize it, or press anywhere outside it and drag to draw a new one. When the result preview looks right, choose a format and press download. On this page the whole job happens in your browser tab, so the photo is never sent to a server and there is nothing to wait for except the final encode.
Switch on circle crop. The box locks to a square, because a circle only fits a square exactly, and both the stage and the result preview show the circular mask. Put the face in the middle of the box, then export as PNG or WebP: the corners outside the circle are saved as real transparency, so the picture sits cleanly on any background. Many sites that show round avatars ask for a square upload and apply the circle themselves, so if a site asks for a square, the plain 1:1 crop is often the better file to give it. The 400 × 400 preset is a sensible size for either.
Type the numbers into the exact pixels fields. X and Y set where the top left corner of the box sits, measured from the picture's left and top edges, and width and height set its size. These are the picture's own pixels rather than the preview's, so a width of 1200 means 1200 pixels in the file. If the final size you need is not a crop of the original pixels, such as exactly 1080 × 1080 from a larger square, crop to the right shape first and then set the output size or pick a preset. The crop decides what is in the picture; the output size decides how many pixels it ends up with.
For Instagram feed posts the long-standing shapes are square 1:1, portrait 4:5, and landscape 1.91:1, while Stories and Reels fill a vertical 9:16 screen. YouTube videos are 16:9, Shorts are 9:16, and the standard YouTube thumbnail is 1280 × 720, which is also 16:9. Every one of those is a preset here except 1.91:1, which you can type as a custom ratio. Platforms do revise their specifications, so check the current guidance before a campaign where the exact shape matters.
Cropping itself does not. It keeps fewer pixels and resamples none of them, so for an ordinary 8-bit sRGB picture a PNG export at the crop's own size keeps the selected pixels unchanged. A browser canvas does change a few things on the way: a wide-gamut photo (Display P3 or Adobe RGB) is converted to sRGB, a 16-bit PNG is saved at 8 bits per channel, and the colour stored in faint, semi-transparent pixels can shift slightly. Two later steps can lower quality more. Saving as JPEG, or as lossy WebP, compresses the result again, and a photo that was already a JPEG goes through a second generation of compression, so keep the quality slider at 90 or above. Resizing the output, and enlarging it in particular, interpolates new pixels from old ones. A crop also simply has fewer pixels than the original, so it shows less detail when it is displayed at the same size.
No. The file is read, decoded, cropped, and encoded entirely inside this browser tab, and the download is created from memory on your own device. There is no server step, so there is no copy to retain or delete. A side effect worth knowing about: the export is drawn onto a fresh canvas, which has nowhere to keep the original's EXIF metadata, so the GPS location, camera model, and capture time are not in the cropped file. Embedded colour profiles are dropped as well, which can shift very saturated colours slightly.
Export as PNG or WebP. Both keep the alpha channel, so a logo or a cut-out with a transparent background stays transparent after cropping, and the result preview sits on a checkerboard so you can see which parts are see-through. JPEG has no alpha channel at all: choose it and every transparent pixel is painted with a background colour you pick before encoding. That is also why JPEG is unavailable while circle crop is on.
Cropping removes the edges of a picture. What is left is a smaller window onto the same pixels, so the subject gets larger in the frame and the shape can change. Resizing keeps the whole picture and changes how many pixels it is made of: the shape stays the same and everything in it gets smaller or larger together. The two are often combined, cropping to the right shape and framing first and then resizing to the size a site asks for, which is why this page has an optional output size after the crop. To scale a whole picture without cutting anything off, use the Image Resizer instead.
Yes. Phones often store a photo in the sensor's own orientation and add an EXIF tag telling software to turn it. The browser applies that tag while decoding, so the picture on the stage is already upright and the crop you draw is the crop you get. The export bakes that orientation into real pixels and carries no tag, so it will not flip sideways in an app that ignores tags. If the picture needs turning beyond that, or a tilted horizon needs levelling, run it through the Image Rotator first.
One file of up to 25 MB, a source picture of up to 40 megapixels, and an export no larger than 8,192 px on its longest side or 40 megapixels in total. They exist because every pixel costs memory in your tab: a decoded picture takes four bytes per pixel, so a 40 megapixel photo is about 160 MB before the crop begins. Past roughly 16.8 megapixels of output the page also warns that some phone browsers hand back a blank image instead of an error. A GIF is cropped as a still picture: one frame is used and the animation is not kept.
Explore the growing collection for calculations, documents, writing, and everyday work.