Update: 08 January 2010
The input images are processed into an output images through a convolution operator.
If both input images do not have the same size, the smaller one should be entirely contained into the biggest, i.e. both dimensions of the smallest should be inferior to the dimensions of the biggest.
The input image #2 plays the role of a convolution kernel, i.e. the
convolution result is spatially shifted by a number of pixels that is
the half if the width and the height of the image #2. This makes the
result does not obey the causality law, but patterns of image #1 are
kept in place.
The result of the convolution processing will be displayed and saved in the current image directory as "convolution.fit". If there were a file with the very same name, its content will be overridden.
The output image is encoded in a floating point format. Its dynamic range is the same as the first input image, i.e. the extrema pixels values are unchanged.