Obligatory "ignore this space" : https://sacoronavirus.co.za

Before I revive old acquiantances in order to drown them, I recall myself to a boy--or a man, if you're feeling generous--who was told a curious tale about a virgin.

Dora Spenlow was a very shallow character. She wasn't interested in financially rewarding applications of mathematics, for instance. Shallow people attract shallow people: quite honestly if you tell me to say mean instead of 'average', I'll tell you about an old man who spent most of his life watching the ocean, thinking that getting children to play with spears was giving them exactly the wrong idea, but knowing that if he said anything he'd have to become an entertainer on overcast days.

Stuff and nonsense, but don't let me interfere with the narrative that gives mathematicians a reason not to lift their head up high and blow their brains out.

But Dora wasn't to be an average mother. Having only the skills she was born with, on an overcast day she sat and talked to her dog. She might have enjoyed reading, but obviously wasn't clever enough to read any but stories of the paranormal.

Deviance from average is necessary for any individual, unless we connect a stop-watch to a trigger; which variance may be the maximum distance or the sum of the absolute difference from average. The choice we make depends on the effect we're trying to achieve.

Before you start thinking I'm a cynical old man who speaks his own language in the face of those who have a bigger-than-average paycheck for the language they use, this is merely a preface for those whose vision goes blurry on sight of the word, half-toning.

An rgb plank can be subdivided by finding the average value of each dimension, for a given sequence of points, and selecting the dimension for which the chosen deviance from the average is greatest. We thus form a tree of divisions, which are cubes, and divide whichever leaf node has the greatest deviance, which may be a different deviance than which is used to choose a dimension.

Generating a palette is a fun task, though not justifiable by any business plan. Dithering produces a better image than half-toning, because it uses paranormal numbers.

Random numbers are paranormal numbers because we insist that they are; we insist that they come to us from outside. There may be such a thing as the best pseudo-random number sequence, but if so, it can only be demonstrated to be the best by our acknowledging that it is a sequence (which gets us looking elsewhere).

Creating a palette-generating algorithm is a random thing to do (we would like to give adjectives to randomness, but that can only demonstrate our ignorance). It can save us from having to give instructions for what can be done automatically, if we choose the right deviance. But we can avoid giving away what hasn't been paid for by spitting out a lossy image when the deviance indicates we need more palette colors than are possible.

Which we must do to avoid complaints about the gradients not being smooth.

And because effects of light keep our eyes open, too.