pamoil reads a Netpbm image as input and does an 'oil
transfer', and writes the same type of Netpbm image as output.
The oil transfer is described in 'Beyond Photography' by
Holzmann, chapter 4, photo 7. It's a sort of localized smearing.
The smearing works like this: First, assume a grayscale image. For
each pixel in the image, pamoil looks at a square neighborhood
around it. pamoil determines what is the most common pixel
intensity in the neighborhood, and puts a pixel of that intensity into
the output in the same position as the input pixel.
For color images, or any arbitrary multi-channel image,
pamoil computes each channel (e.g. red, green, and blue)
separately the same way as the grayscale case above.
At the edges of the image, where the regular neighborhood would run
off the edge of the image, pamoil uses a clipped neighborhood.
OPTIONS
-nsize
This is the size of the neighborhood used in the smearing. The
neighborhood is this many pixels in all four directions.