This is a basic, text comes in, piet image comes out, converter. You supply a text file, and a piet program which outputs the given text is output as an image.
The output image is always square, for simplicities sake. The side of the square, henceforth referred to as the image size, can be either passed in as a parameter, or the converter can try to find a close fit for the minimum image size. It is not guaranteed that the guessed image size will be large enough to accommodate the whole piet program, but it hasn't happened to me yet, and it shouldn't happen (at least I tried to keep it from happening).