In mt5.ucl there's a coloring algorithm called "Convolution" (this is really just a shell for the ConvolutionColoring class - but you don't need to know that). Convolution is (and I quote) "an image processing technique in which each pixel is altered by some function of the surrounding pixels; different functions produce a variety of effects."

I've written convolution coloring algorithms before (and you might have even tried to use them) but UF5 allows me to take it to a whole new level. The new Convolution coloring lets you try the effect on any fractal formula and any coloring algorithm (as long as there are class versions available).

Try this... Start a new fractal and load the Pixel formula in Standard.ufm. Go to the Outside tab and browse for Convolution in mt5.ucl. You'll see a parameter called "Kernel" and it will read "Center" and next to that there'll be a browse button. Click on the browse button and try the other kernels to see what effect they have. For now these are the only effects available but more can be written (if you want to write your own kernels then the programmers list would be the place to ask if you need help).

Now a quick run through the parameters...

• Sample distance. It's best to think of this as a kind of "strength" parameter. With the Edge Detect kernel larger values will make the lines thicker, and with the Soften kernel they will make things more blurry. This parameter is adjusted for the zoom, so the effect will remain reasonably constant.


• I've already gone over the Kernel parameter.


• Fractal Formula lets you select which formula to use. Look through the various *.ulb files to see what's available. The parameters below this are the same as what you'd normally see in the Formula tab. One thing to note is that this coloring uses the Maximum Iterations parameter on the formula tab for the selected formula.


• Coloring Algorithm lets you select the coloring algorithm to use (equivalent to the normal "outside" coloring).


• Inside color. You can't select an "Inside" coloring for the formula (that would be just too slow), but this allows to select the color that inside areas appear.


• Solid color. This is the color used for "Solid: areas if the coloring algorithm defines them.


• Opacity. There are three Opacity modes. I've just realized that "None" means the opposite of what it says: all pixels remain fully opaque. "Gradient" means that the opacity of the gradient is used. This includes the opacity of the Inside and Solid colors. "Luminance" means that the luminance of the pixel is used to define opacity. In this mode a black pixel will be fully transparent and a white pixel will be fully opaque.

I know that's a lot to take in, so if your interested in playing with this formula you might file this message away for later reference.

The UPR below uses one layer of Convolution. You'll note that I've used the Edge Detect kernel in "Luminance" opacity mode as a mask to put black lines over the image.

Sample Parameters