Little squares.

Computer generated pictures could be pixel-first or object-first. Yes, I made up those terms. Fractals are pixel-first. Pick a pixel, run the fractal calculation for the pixel to get a color, color the pixel, then move on to the next pixel.

This image consists of many small squares, and is drawn object-first. Pick a square, compute which pixels are on the square, color those pixels with the color of the selected square, and then move on to the next square.

I started down this Port/Refactor path figuring I would just focus on programming for a few weeks. That has turned into half a year, with no end in sight. (And now less time to work on it.)

With a little effort I could turn each of these diversion into different series. But when I start thinking about it, I immediately notice all of the programming things that are not yet done. Or worse, the program features that are broke or disabled while working on other fixes. The program runs just well enough to test new features.

So, please indulge me as I post these "experiments", and drop them into the "Refactor" bucket.

I came across a few more pieces from when I was experimenting with random textures and surfaces. After some minor clean up, they are ready to publish.

A couple more without color blending.

Jungle. A lot of overlapping colors.

sin waves in blue and brown

Making the sin waves obvious.

Hot palette

Diamond shapes

Holes

There are two frequencies involved here. Call them x and y. When one is much larger than the other you get horizontal or vertical waves. If they are nearly equal you can see both types of waves.

Vertical waves in two colors. Playing with the direction and density.

In this variation the colors are not blended. Where they would otherwise overlap, one color is selected, rather than trying to blend them.

For the nerds out there, the bug was one of those kind of bugs. Internally I represent each color channel by a (continuous) floating point value between 0 and 1. The graphics hardware and the bitmap file format uses (discrete) byte values between 0 and 255. So one must multiply or divide by 256, depending on direction, for conversion. At one point I wrote 255 instead of 256. I looked at that number many times and it did not occur to me that it was wrong. The small difference is barely noticeable.

Small tweaks to the parameters, and added a black stripe to the vertical palette.

I had put in several parameters to fine tune the results. While the results were close to my intent, the parameters did not work as I expected. Sometimes I liked the unintended results. But mostly it was frustration trying to get better control over the finer details.