Interpolation filter
James Lambie 2016-12-22T05:50:53Z
193 words
1 min read

URL parameter: filter, f

This parameter allows you to specify the interpolation method to use when resizing images. If not specified, the default is lanczos.

When reducing the size of an image (or downsampling), some image data is simply discarded. However when increasing image dimensions, the image is expanded and gaps must be “filled in”. Each interpolation filter uses a different algorithm for determining how to fill the gaps.

Possible values:

  • lanczos (default)
  • nearest-neighbor
  • linear
  • cubic
  • grid
  • moving-average




The simplest approach to interpolation. Rather than calculating an average value by some weighting criteria or generating an intermediate value based on complicated rules, this method simply determines the “nearest” neighbouring pixel, and assumes the intensity value of it.


Considers the closest two pixels and takes a weighted average to arrive at its final interpolated value. Results in a much smoother image than nearest-neighbor.


Images resampled with cubic interpolation are smoother and have fewer interpolation artifacts, but processing is slower than with linear or nearest-neighbor.


Tends to reduce aliasing artifacts and preserve sharp edges. It has been considered the “best compromise” among several simple filters for this purpose.

See something to improve? Edit in Github.