One thing to note is that I didn't quite use the same values in the GIF that I did in the first posting. One of the tricky/annoying/useful aspects of the algorithms in Wilbur is that they're sensitive to scale because they work in pixels, not abstract units. The precipiton erosion feature, for example, always generates channels that are some number of pixels wide; the morphological operations are always operate on an area 3 pixels wide; the incise flow operation deals with flow across image samples (pixels); and the blurs are defined in something that works out to be pixels. How large the effects of each of these activities appear is relative to the overall image is a function of the size of the image. The information in the GIF uses a 256x256 image, while the other one is a 512x512 image. Note the subtle differences. As a benefit, the smaller image processes much faster than the larger one (some of those algorithms are O(n*n*n), meaning that an image half the size may take 1/8 the time).
A loose version of the classic definition of a fractal deals with the notion of a motif repeated across scales. An entertaining use of Wilbur is to do the bulk processing of a surface at relatively small size (256x256 is a good starting size), scale the surface up by a factor of 2 (or thereabouts) and then apply processing for higher details. A common processing operation in Wilbur involves basins fills and incise flow. Both of these operations are fairly slow, but the incise flow is especially so. By roughing in the shapes at lower resolution, it's fairly quick to get something done. The attached crop of a 4kx4k image done via scale-and-add shows what 10 minutes of processing will do (I forgot to punch out the crater, so it's just a mountain). Trying to get this same level of effect by starting at 4kx4k usually takes me about an hour.
Reply With Quote