How It Works

What happens inside your browser

When you add an image, the browser decodes the file into a bitmap in memory and draws a scaled preview of it onto a canvas element. Every adjustment you make — moving the box, dragging a handle, zooming, rotating — is stored as a set of coordinates on that preview. When you press Apply Crop, the coordinates are mapped from the on-screen view back to the original pixels, and the browser composites the selected region onto a fresh canvas at full resolution. The exported file is encoded on your device from that canvas.

Because the entire pipeline runs in your browser, there is no upload step to secure. The site has no upload endpoint: no form posts your file anywhere, and no background script sends image data to a server. You can open your browser's network tools while cropping and confirm that nothing leaves the page — the only requests are the ones that fetch the site's own HTML, CSS and JavaScript when it loads.

The same applies after you finish. Your image lives in the browser tab's memory while you work; close the tab and it is gone — there is no cache to purge and no account holding a copy. The privacy policy documents the full data practices behind this design.

Smart crop suggestions, honestly explained

The Smart crop button is the tool's "AI" feature, and it is worth being exact about what that means. When you press it, the page analyzes visual energy across your image — where edges, texture and detail are concentrated — using classic image-analysis techniques that run on your device. It then proposes the crop window that keeps the most active region in frame, biased slightly toward the center for a natural balance.

This is not machine learning. There is no neural network, no model download and no training data involved — the suggestion comes from deterministic mathematics applied to your pixels, which is why it works instantly and behaves identically on the same image every time. What it replaces is the guesswork of blind center-cropping: instead of cutting straight down the middle, the suggestion follows the detail in your specific picture.

The suggestion is a starting point, never a verdict. You are expected to fine-tune it — nudge the box, adjust the handles or ignore it entirely and crop by hand. And because the analysis happens locally, it is private by construction: computing a suggestion sends nothing anywhere, adds no network waiting, and works the same way whether you are on a fast connection or offline once the page has loaded.

Tips for better crops

A few habits make the difference between a crop that works and one that needs redoing:

• Lock the aspect ratio first. Pick your target ratio before fine-tuning, so the box already has the shape you need and every adjustment stays within it.
• Put the subject near a third line. Off-center framing usually reads more naturally than dead-center — imagine the frame split into thirds and place your subject on or near one of the lines.
• Leave headroom for circular avatars. Platforms often round 1:1 crops into circles or add overlays, so keep a margin above heads and around edges instead of cropping tight.
• Crop once, from the original. Repeatedly cropping saved exports compounds compression loss — this tool always cuts from the full-resolution source, so start from your best file.
• Match the format to the content. Export screenshots and anything with text as PNG to keep edges crisp; choose JPEG or WebP for photographs where smaller files matter.
• Check the pixel label before applying. The size label on the crop box shows the exact output dimensions — confirm them before pressing Apply Crop and the download will match your target.