Randomness in CSS using trigonometry

hypersphere.blog

88 points·thunderbong··8 comments

Title already makes me love the article.

Whenever I got back to how fun it is to plot 2D functions, I soon started to try creating freaky images with parts of seeming chaos.

Like combining trigonometric, algebraic and discontinous functions such as the floor function.

Unfortunately, https://en.m.wikipedia.org/wiki/Thomae%27s_function

requires branching and a complex discontonuous and uncomputable (?) function in the definition :)

IIUC this gives a random distribution between 0.0 and 1.0 but it is not a uniform distribution. There are more results near 0.0 and 1.0 than near 0.5. If you use this function twice to draw random black points on a white square, the corners will look darker than the center.

If you replaced the sine function with a sawtooth wave (n - floor[n]), you would have continuous, uniformly distributed "pseudorandom" numbers from this method.

I think the round() function in CSS is still experimental, but you may be able to get a similar effect by intentionally underflowing a floating-point calculation.

Arcsin(sin(x)) would also work, or arccos(sin(x)), but you would probably have to make an approximation of arcsin/arccos for yourself.

I agree, but for a continuous function you need abs((n - floor[n])-1/2)*2 https://www.wolframalpha.com/input?i=abs%28%28n+-+floor%5Bn%...

I like the article, but what irks me is the use of the word "pseudorandomness," which has a technical definition in TCS.

Reply on news.ycombinator.com