logo
 

Resampling filters

 

Resampling is a technique that resizes an image using special filters.

The resample filter (or resize algorithm) determines how pixels in a picture are affected when it is resized. Using the right resample filter will keep your enlargements looking great. Some filters add a blurring effect while others maintain the hard edges between different areas of a picture.

The filter that generally gives the best results is "Lanczos3". However, depending on your thumbnail, you may get better results (smaller file size or higher image quality) by trying out one of the other filters.

You can apply several resample filters to your image:

Triangle

This filter produces sharp transition lines and is relatively fast. It produces reasonably good results with a little smoothing for both reduction and magnification.

Hermite

The Hermite method creates noticeable smoothing but maintains a good degree of sharpness. It is relatively fast.

Bell

This filter is also relatively fast and produces smooth image, bordering on soft.

B-spline

This method does not produce sharp transitions and may cause excessive blurring. It has the advantage of dampening noise and JPEG artifacts. B-spline is one of the slower filters.

Lanczos3

This is a slow method but it usually produces the sharpest images. Under certain conditions, it may introduce some ringing patterns and emphasize JPEG artifacts.

Mitchell

Resizing with this algorithm produces no sharp transitions and tends to be a good compromise between the "ringing" effect of Lanczos and "blurring" of other methods. The Mitchell algorithm produces good results when enlarging pictures. It is also one of the slower filters.

Nearest

This filter is recommended for small icons or lines/outlines. It works best with illustrations containing non-anti-aliased edges.

Linear/Fast Linear

This is a reasonably fast method that does not emphasize jagged edges, but it does tend to blur the image slightly.

None

The image is simply stretched to the new size. This is the fastest method but produces the least smoothing.

More info...