Images that have noisy, low contrast, or blurry boundaries can end up with jagged results. Use the scalpel and the menu to deal with this:

  1. Use the scalpel to specify the boundary in sections with high noise or low contrast.

  2. Experiment with different levels of smoothing until the jaggedness is under control.

  3. Recover any lost corners by adding green / red corner marks.

  4. For severe cases of halo, where the background bleeds into the foreground, you can increase the feathering or offset the boundary to crop the halo out.

Clipping Magic features automatic corner detection to keep your result nice and crisp even when there's a lot of smoothing applied. It's on by default, but you can turn it off if you prefer.

Corners:

tutorial image
Original + Marks
Result

Corners:

tutorial image
Original + Marks
Result

You normally want to use the lowest smoothness setting that gets rid of the jaggedness sufficiently. This is to preserve the feel of the outline and to avoid rounding out undetected corners unnecessarily.

Adaptive Smoothing uses a smoothing level that adapts to the local character of the foreground, so sharp edges are followed more closely, and fuzzy edges are smoothed out more.

Fixed Smoothing uses the user-specified smoothing level along the entire boundary.

Adaptive Smoothing and a setting of 1 usually achieves a nice foreground boundary. However, the best setting will depend on your particular image - here is a sample case as illustration:

You may need to turn off Corner Detection to allow the smoothing to take full effect as the corners may otherwise make the result faceted.

Smoothing Level: 1

tutorial image
Original + Marks
Result

Smoothing Level: 0

tutorial image
Original + Marks
Result

Smoothing Level: 2

tutorial image
Original + Marks
Result

Smoothing Level: 5

tutorial image
Original + Marks
Result

Some corners will inevitably go undetected by the algorithm, or you may have turned off Corner Detection. And sometimes the image requires a smoothing setting that is so high that the missed corners get rounded out. You can reclaim these corners by adding key green / red marks to force the clipping path to respect the corner:

Corner Marks

Corners:

Smoothing Level: 5

tutorial image
Original + Marks
Result
×

×

tutorial image
Original + Marks
Result

The feathering radius controls how many pixels it takes to transition from fully opaque foreground to fully transparent background. A small radius creates a sharp edge that is suitable for hard objects; a large radius gives an appearance of softness that works well for certain textiles and other soft or fuzzy objects.

While the algorithm normally produces clean foregrounds that are not polluted by so called 'halos' (background bleeding into the foreground), increasing the feathering radius can help in severe cases.

Adaptive Feathering uses a feathering radius adapts to the local character of the foreground, so sharp edges stay sharp and fuzzy edges stay fuzzy. Works great on shots with varying levels of focus blur along the foreground edges!

Fixed Feathering uses the user-specified feathering radius along the entire boundary.

Adaptive Feathering and a setting of 1px usually achieves a very faithful reconstruction of the foreground boundary, and is the recommended setting.

Here is a sample case as illustration:

Feathering Radius: 1px

tutorial image
Original + Marks
Result

Feathering Radius: 3px

tutorial image
Original + Marks
Result

Feathering Radius: 6px

tutorial image
Original + Marks
Result

The algorithm normally produces clean foregrounds that are not polluted by so called 'halos' where the background bleeds into the foreground.

However, in severe cases, or when a very tight feathering radius is desired, you can crop out any halos by offsetting the boundary inwards.

The best setting will depend on your particular image and intended use - here is a sample case as illustration:

Offset: 0.0px

tutorial image
Original + Marks
Result
Halo

×

Offset: 2.0px

tutorial image
Original + Marks
Result
No Halo


Pre-Crop

Your image size exceeds your active size limit. For best results, please crop the image to the portion you wish to clip.


Size Limit


Original Image

Size:
Aspect Ratio:
Megapixels:

Cropped Image

Size:
Aspect Ratio:
Megapixels:
Cropped image exceeds size limit and will be scaled to fit.
Size limit met, full resolution preserved.

Javascript Error :-(


Please help us fix this issue!