Crop without guessing the frame.

Manual is best when you have a specific composition in mind and want pixel-precise control — golden-ratio framing, headline space, asymmetric crops. Smart crop is best when you're fanning one source to multiple aspect ratios and need the focal subject (face, product, headline) to stay in frame across every variant. They share the same overlay, so you can switch between them mid-edit.

A focus point — centre, top, bottom, left, right — anchors a region of the source image. When the target aspect changes, smart crop slides the crop window so that anchor stays inside the frame, instead of cutting through it. It's a deterministic heuristic, not a remote ML call: nothing leaves the browser.

Yes. Any crop you make — manual rectangle plus aspect ratio plus focus point — can be saved as a preset and re-applied to a different source with one click. Custom presets sit alongside the published twelve in the same shelf.

Always smaller, never larger. A crop discards pixels outside the overlay; the encoder writes a smaller file by definition. If you also re-encode (PNG → WebP, say) you'll see a second round of savings — the studio shows you the new file size before export.

Yes. Drop multiple files into batch mode, set a smart-crop focus point and aspect ratio once, and the queue applies the same crop logic to every input. Pair with Convert in a recipe to ship the whole folder as a single zip.

No. Both manual crop and smart crop run entirely on the Canvas API in your tab — same engine the rest of the studio uses. The source bytes stay on-device; only the cropped output you save touches your disk. Refreshing the page clears state.