There are multiple ways of specifying the size and how that size is achieved.


Cropping is the better option if what you need is a reduction of image size where some of the edges may be lost in exchange for otherwise keeping the image quality identical. Resizer will crop by the required amounts from all four edges. If for example height is specified as 90% and width as 80%, the cropped image will have 5% removed from the top, 5% from the bottom, 10% from the left and 10% from the right. If percentages are specified, they must be between 0 and 100. If a specific size is specified, all images, regardless of their original size, will be cropped in the symmetrical way described above where the output image is smaller than the input image. Otherwise a black border will appear to fill in the additional space.


Expand/Squeeze is the better option if you do not wish to lose any part of the image and do not want a border. There are, however, multiple ways of resampling the image to change in size, both for enlargement and reduction. Use the Additional Settings button at the bottom of the main window to specify the required resampling algorithm. Plenty of information can be found about these algorithms on the internet, but very briefly:

  • Nearest (Neighbour) is a fast algorithm which preserves details but can result in jaggedness.
  • Bilinear is an algorithm which offers less jaggedness than Nearest (Neighbour) by softening edges, but some detail can be lost as a result.
  • Bicubic is a popular algorithm when speed is not an issue thanks to the appeal of its smoothness
  • AntiAlias refers to the famous Lanczos algorithm considered a good option if you do not have specific requirements as it is a popular compromise between competing considerations


In order to retain the aspect ratio make sure that when % is chosen and you opt for Expand/Squeeze you select the same percentage for width and height. Obviously it’s not possible to do the same for the case of pixels unless all your input images have the same aspect ratio.


Back to ImageShell Home Page