The crop decides what the photo is about
A wide shot squeezed into a square loses its subject. Drop a photo, drag the frame over the part that matters, pick the size the app asks for, and download it at exactly those pixels - no stretching, no uploading.
The Image Cropper trims a photo to a ratio such as 1:1, 4:5 or 16:9, or to exact pixel dimensions, with a preview of what the crop removes. Cropping happens on your device, and you save the result in the format you need.
Resizing changes the shape. Cropping chooses it.
Two operations get confused constantly. Resizing keeps the whole picture and changes how many pixels it is made of - the shape stays the same. Cropping keeps the pixel count roughly the same and throws away everything outside a frame - the shape changes, the content changes, and the subject gets bigger without being distorted.
When an app asks for 1080 by 1350 pixels, it is describing a portrait frame. A landscape photo forced into that frame either stretches the faces or adds thick bars. A crop keeps the proportions honest and simply says: this part of the scene is the photo now. For a portrait, a product shot or a cover image, that is almost always the better answer.
The frame is the whole decision. Move it left and the subject sits on the right third; make it taller and the horizon moves down. That is why this tool shows the shaded, discarded area while you drag: you are not only choosing what to keep, you are choosing what to lose, and the second one is the part people notice later.
The sizes the big platforms actually ask for
| Where | Ratio | Pixels | What gets covered |
|---|---|---|---|
| Instagram feed | 4:5 | 1080 × 1350 | nothing, but the caption sits below |
| Reels, Shorts, TikTok | 9:16 | 1080 × 1920 | top and bottom bars hold buttons and text |
| YouTube thumbnail | 16:9 | 1280 × 720 | the duration badge, bottom right |
| YouTube banner | 16:9 | 2560 × 1440 | only the middle 1546 × 423 is safe on a phone |
| Facebook cover | 2.7:1 | 851 × 315 | the profile picture overlaps the left side |
| X header | 3:1 | 1500 × 500 | the avatar covers the lower left corner |
| LinkedIn banner | 4:1 | 1584 × 396 | the profile card sits over the lower left |
| Profile picture | 1:1 | 400 × 400 | everything outside the circle |
Three moves and the framing is done
Drop the photo
Any JPG, PNG, WebP or GIF frame. The frame opens at the largest size the chosen ratio allows.
Drag the frame
Corners resize it, the inside moves it. With a fixed ratio the shape is locked, so the crop stays exact.
Download at the target size
The result is drawn at the platform's pixel size, not at the size of the preview - sharp on any screen.
Three things that decide whether the crop works
Faces first. If the frame cuts a forehead or a chin, the photo looks like an accident; a bit of room above the head costs almost nothing and reads as deliberate. Hands and products matter next - a held object that is half cropped looks broken rather than composed.
The rule of thirds in the overlay is there for a reason: a subject on one of the lines, looking into the empty space, feels balanced. A subject dead centre is fine too, but it needs symmetry around it. What almost never works is a subject at the very edge with the empty side behind them.
Finally, check the edges for things you did not notice: a chair leg, a cable, half a person. A crop that removes a distraction is worth more than one that keeps a beautiful background.
Cropping is right when something can be lost: an empty sky, a repeat of the same wall, the outer thirds of a wide scene. Padding is right when nothing can be lost: a group photo where everyone matters, a logo with text at both ends, a diagram that fills the frame.
The compromise most social tools use is a blurred copy of the same photo behind a padded frame - it keeps every pixel and fills the empty space with colour that matches. That is not a crop, it is a background, and it is a different tool.
One rule of thumb: if the crop would remove a person, pad; if it would remove wallpaper, crop.
Cropping cannot invent pixels. If a 3000 pixel wide photo is cropped to a quarter of its width and then exported at 1080 pixels, the result is an upscale: soft, with visible artefacts on edges. The tool still produces the exact size the platform wants, but the sharpness comes from the source.
The safe habit is to crop from a photo at least as large as the target in both directions. A 1080 by 1350 export wants a source of roughly 1080 by 1350 inside the frame - which most phone photos comfortably have.
If the source is smaller, export at the smaller size and let the platform scale it up: their resampling is generally better than a second round of lossy compression on your side.
Questions about framing
Does cropping reduce the quality of my photo?expand_more
Why is the exported file bigger than the crop I see?expand_more
Can I crop several photos at once?expand_more
What happens to transparent PNGs?expand_more
Can it find the face automatically?expand_more
Why is the frame locked to a ratio?expand_more
One frame, one download
Drag the frame, pick the size the platform wants, take the file.
fact_checkQuick facts
- A square crop of a 4:3 photo keeps 75% of its width, because the square is as wide as the frame is tall.
- The pixels that survive a crop are untouched, but saving the result as JPEG re-encodes them once.
- JPEG splits the picture into 8×8 pixel blocks and stores a discrete cosine transform of each block, which is where its block-shaped artefacts come from.menu_bookITU-T
- JPEG has no alpha channel, so transparency always becomes white or black when a picture is saved as JPG.menu_bookITU-T
- PNG compresses with DEFLATE, so decoding always returns exactly the pixels that were saved.menu_bookW3C
helpQuestions people ask
How do I crop a photo to a square?expand_more
Does cropping reduce image quality?expand_more
What is the best crop for a profile picture?expand_more
hubWhat this page is about
- JPEGopen_in_new
A lossy still-image format standardised as ITU-T T.81 in 1992 and still the default format of most cameras and phones.
- PNGopen_in_new
A lossless image format defined by the W3C with an 8-bit alpha channel and gamma information, designed for the web in 1996.
- ICC colour profileopen_in_new
A file that describes how the colours of an image map to a device, defined by the International Color Consortium.
- Canvasopen_in_new
The HTML element that gives a page a pixel buffer to draw on and export as PNG, JPEG or WebP.
library_booksOfficial sources cited on this page
Continue with the same file
Each of these opens what the tool above produces. Use “Use in another tool” under the result to carry the file over — it moves from tab to tab inside your browser, never to a server.
Switched-off tools are not listed here: only what works right now.