Image Trim (Auto-Crop Transparent / Solid Borders)

How to use

1. Drag an image onto the drop zone or click to choose a file.
2. Pick the background mode — Auto works for most images (samples corners).
3. Bump tolerance up if the border isn't a perfectly uniform color (JPEG noise, near-white scans).
4. Add padding if you want a clean margin around the content.
5. Choose export format and click Download — savings vs original area is shown alongside the trimmed preview.

Frequently asked questions

How does Auto mode pick the background color?

It samples the four corner pixels. If any corner is transparent (alpha ≤ 8), it switches to transparent mode. Otherwise it takes the median R, median G, and median B across the four corners as the target color. The median is robust to one weird corner — if three corners are white and the fourth happens to fall on actual content, the median still reports white. For pathological cases (content touching multiple corners), use one of the explicit modes instead.

What does tolerance do?

A pixel counts as background if it's within `tolerance` of the target color on every channel (max channel difference ≤ tolerance). For transparent mode, tolerance is mapped to an alpha threshold (alpha ≤ tolerance × 2.55 counts as background). At tolerance 0 only exact matches are trimmed; at 10-20 you'll trim near-white scans and JPEG-noisy borders; at 50+ you'll start eating into real content. Push it up until the trim looks right.

Why does the result have a checkered background?

That's the standard transparency indicator — wherever you see the checkerboard pattern, the image has alpha = 0 (fully transparent). It's only there in the preview; the actual file has true transparency. If you export as JPEG (which can't store transparency), the JPEG background color you pick fills those areas instead.

What if the whole image is the background color?

You'll see 'no content found' instead of an output. That happens when nothing in the image differs from the target color by more than the tolerance — usually because you picked the wrong mode (e.g. 'transparent' on a flattened JPEG, or 'black' on a light image). Switch mode or raise tolerance.

Does padding ever exceed the original image bounds?

No. Padding is clamped to the original image edges. If your content was already flush against one side, padding adds extra space only on the sides where there was room. The output dimensions are always inside the source dimensions — this tool only crops, it never expands the canvas. Use the Image Padding tool if you want to add canvas beyond the original.

Related tools