Image Compressor
Shrink JPG, PNG and WebP with a quality slider and live size readout.
Image & Photo
Crop with draggable handles, aspect presets, rotate and flip.
Cropping removes pixels from the edges of the frame and leaves the remaining pixels at their original scale. Resizing keeps the entire frame and changes the scale of every pixel in it.
That distinction has a practical consequence: cropping never softens the image, because the surviving pixels are untouched. Resizing down is always a resampling operation, and resizing up always invents detail.
When you need a specific shape, crop. When you need a specific size, resize. When you need both — a 1080×1080 Instagram post from a 4:3 photo — crop to square first, then resize the square down. Doing it in that order means the resample happens once, on already-correctly-framed content.
Aspect ratio is the shape; pixel dimensions are the size. Crop to the shape, then resize to the size.
| Ratio | Shape | Used for |
|---|---|---|
| 1:1 | Square | Instagram posts, profile photos, app icons, avatars |
| 4:5 | Portrait | Instagram portrait — the largest feed footprint |
| 9:16 | Tall | Stories, Reels, TikTok, YouTube Shorts |
| 16:9 | Widescreen | YouTube thumbnails, presentation slides, video |
| 1.91:1 | Wide banner | Open Graph cards, Facebook and LinkedIn link previews |
| 2:3 | Tall portrait | Pinterest pins, book covers, 4×6 prints |
| 3:2 | Classic photo | 35mm film ratio, most DSLR sensors, 6×4 prints |
| 4:3 | Traditional | Older sensors, iPad screens, print documents |
You crop to a 1:1 square and let the platform apply the circle. Every service that shows round avatars — Slack, Google, LinkedIn, Discord, GitHub — accepts a square image and masks it in CSS at display time.
Uploading an image with actual transparent corners is usually worse, because some surfaces render avatars as squares or rounded rectangles and the transparency then shows as a white or checkerboard notch.
The one thing to get right is the safe area. A circular mask on a square image cuts off about 21% of the total area at the corners. Keep the subject’s head inside the central circle, with clearance, or the crop will clip an ear.
Yes, roughly in proportion to the area removed. Cropping a 4000×3000 image to a 3000×3000 square removes 25% of the pixels and typically removes a similar share of the bytes.
It is not exactly proportional, because compressed size depends on content complexity as well as pixel count. Crop away a plain sky and you remove many pixels but few bytes; crop away a detailed crowd and the file shrinks disproportionately.
No. The surviving pixels are copied unchanged. Quality only changes if you also re-encode to a lossy format, which is why the tool offers PNG output for lossless results.
Yes. The width and height fields accept exact values, and the crop box snaps to them. The readout shows the live pixel dimensions of the current selection.
Within the tool, yes — the original stays loaded until you drop a different file, so you can reset and start again. Once downloaded, the cropped file is a new file; your original on disk is never modified.
Drop the screenshot, drag the handles to exclude the toolbar and status bar, and export as PNG to keep the text sharp.
It supports 90-degree steps and flips. Arbitrary-angle rotation requires resampling every pixel and leaves empty corners that need filling or a secondary crop, so it is out of scope here.
Not in one action — crop position is inherently per-image. For uniform output dimensions across a batch, use the Image Resizer instead.
No. The image is drawn to a canvas in this page and the crop is a canvas operation. Nothing leaves your device.