Красочные нейроны
Рандомная генерация красочных нейронных связей на canvas
HTML
<canvas></canvas>
<!-- Click to generate a new image -->
CSS
html, body {
overflow: hidden;
touch-action: none;
content-zooming: none;
position: absolute;
margin: 0;
padding: 0;
width: 100%;
height: 100%;
background: #000;
}
canvas {
position: absolute;
width: 100%;
height: 100%;
user-select: none;
cursor: pointer;
}
JS
// adapted from https://www.contextfreeart.org/gallery2/index.html#design/299
// Colourful neurons by krajzega, May 30th, 2006
"use strict";
const cfa = {
canvas: document.querySelector("canvas"),
queue: [],
identity: [1, 0, 0, 1, 0, 0, 0, 1, 1, 1],
init(background) {
this.ctx = this.canvas.getContext("2d");
this.width = this.canvas.width = this.canvas.offsetWidth * 2;
this.height = this.canvas.height = this.canvas.offsetHeight * 2;
this.ctx.fillStyle = background;
this.ctx.fillRect(0, 0, this.width, this.height);
this.rect = [0, 0, 0, 0];
this.queue.length = 0;
this.scale = 500;
this.minSize = 20;
this.iter = this.run();
},
ajustments: {
x(m, v) {
m[4] += v * m[0];
m[5] += v * m[1];
},
y(m, v) {
m[4] += v * m[2];
m[5] += v * m[3];
},
rotate(m, v) {
const cos = Math.cos(Math.PI * v / 180);
const sin = Math.sin(Math.PI * v / 180);
const r00 = cos * m[0] + sin * m[2];
const r01 = cos * m[1] + sin * m[3];
m[2] = cos * m[2] - sin * m[0];
m[3] = cos * m[3] - sin * m[1];
m[0] = r00;
m[1] = r01;
},
scale(m, v) {
let x, y;
if (Array.isArray(v)) {
x = v[0];
y = v[1];
} else {
x = v;
y = x;
}
m[0] *= x;
m[1] *= x;
m[2] *= y;
m[3] *= y;
},
hue(m, v) {
m[6] += v;
m[6] %= 360;
},
sat(m, v) {
this.adjustColor(m, v, 7);
},
bri(m, v) {
this.adjustColor(m, v, 8);
},
alpha(m, v) {
this.adjustColor(m, v, 9);
},
adjustColor(m, v, p) {
let c = m[p] + Math.abs(v) * ((v < 0 ? 0 : 1) - m[p]);
if (c < 0) c = 0;
else if (c > 1) c = 1;
m[p] = c;
}
},
adjust(s, p) {
const m = s.slice();
for (const c in p) {
this.ajustments[c](m, p[c]);
}
return m;
},
setTransform(s) {
this.ctx.setTransform(
-this.scale * s[0],
this.scale * s[1],
this.scale * s[2],
-this.scale * s[3],
this.scale * s[4] + this.offsetX,
-this.scale * s[5] + this.offsetY
);
},
CIRCLE(s, p) {
s = this.adjust(s, p);
cfa.queue.push([s, "circle"]);
cfa.boundingRect(s);
},
SQUARE(s, p) {
s = this.adjust(s, p);
cfa.queue.push([s, "square"]);
cfa.boundingRect(s);
},
circle(s) {
this.setTransform(s);
this.fillStyle(s);
this.ctx.beginPath();
this.ctx.arc(0, 0, 0.5, 0, 2 * Math.PI);
this.ctx.fill();
},
square(s) {
this.setTransform(s);
this.fillStyle(s);
this.ctx.fillRect(-0.5, -0.5, 1, 1);
},
fillStyle(s) {
const light = (2 - s[7]) * (s[8] * 0.5);
const sat = light <= 1 ? s[7] * s[8] / light : s[7] * s[8] / (2 - light);
this.ctx.fillStyle = `hsla(${s[6]},${sat * 100}%,${light * 100}%,${s[9]})`;
},
boundingRect(s) {
const x = s[4] * this.scale;
const y = s[5] * this.scale;
if (x < this.rect[0]) this.rect[0] = x;
else if (x > this.rect[1]) this.rect[1] = x;
if (y < this.rect[2]) this.rect[2] = y;
else if (y > this.rect[3]) this.rect[3] = y;
},
center(s) {
const br = this.rect;
const scale =
Math.min(this.width / (br[1] - br[0]), this.height / (br[3] - br[2])) * s;
this.scale *= scale;
this.offsetX = this.width * 0.5 - (br[0] + br[1]) * 0.5 * scale;
this.offsetY = this.height * 0.5 + (br[3] + br[2]) * 0.5 * scale;
},
execRule(s, p, rules) {
s = this.adjust(s, p);
if (
Math.abs(s[1]) * this.scale < this.minSize &&
Math.abs(s[3]) * this.scale < this.minSize
)
return;
let totalWeight = 0;
for (const rule of rules) totalWeight += rule.weight || 1.0;
let weight = 0,
rnd = Math.random() * totalWeight;
for (const rule of rules) {
weight += rule.weight || 1.0;
if (rnd <= weight) {
rule.exec(s);
break;
}
}
},
*run() {
let s = 0;
for (const draw of this.queue) {
this[draw[1]](draw[0]);
if (s++ > 10) {
s = 0;
yield requestAnimationFrame(_ => this.iter.next());
}
}
},
render() {
this.iter.return();
this.iter = this.run();
this.iter.next();
}
};
["click", "touchdown"].forEach(event => {
document.addEventListener(event, e => start(), false);
});
/*
o
\_/\o
( Oo) \|/
(_=-) .===O- ~~Z~A~P~~ -O-
/ \_/U' /|\
|| |_/
\\ |
{K ||
| PP
| ||
(__\\
*/
const neuron = (s, p) => {
cfa.execRule(s, p, [
{
exec(s) {
cfa.CIRCLE(s, { scale: 1.02 });
cfa.CIRCLE(s, { scale: 0.8, hue: 20, sat: -1, alpha: -0.5 });
synapse(s, { rotate: 30, y: 0.54, scale: 0.9, hue: 1 });
synapse(s, { rotate: -30, y: 0.54, scale: 0.9, hue: 1 });
synapse(s, { rotate: -110, y: 0.54, scale: 0.9, hue: 1 });
}
}
]);
};
const synapse = (s, p) => {
cfa.execRule(s, p, [
{
weight: 20,
exec(s) {
cfa.SQUARE(s, { y: 0.2, scale: [0.06, 0.4] });
synapse(s, { y: 0.4, rotate: 10, hue: -10 });
}
},
{
weight: 20,
exec(s) {
cfa.SQUARE(s, { y: 0.2, scale: [0.06, 0.4] });
synapse(s, { y: 0.4, rotate: -10, hue: 10 });
}
},
{
weight: 6,
exec(s) {
endsynapse(s);
}
},
{
weight: 4,
exec(s) {
neuron(s, { scale: 0.9, hue: 2 });
}
}
]);
};
const endsynapse = (s, p) => {
cfa.execRule(s, p, [
{
weight: 20,
exec(s) {
cfa.SQUARE(s, { y: 0.2, scale: [0.06, 0.4] });
endsynapse(s, { y: 0.4, rotate: 10, scale: 0.9, alpha: -0.2 });
}
},
{
weight: 20,
exec(s) {
cfa.SQUARE(s, { y: 0.2, scale: [0.06, 0.4] });
endsynapse(s, { y: 0.4, rotate: -10, scale: 0.9, alpha: -0.2 });
}
}
]);
};
const start = () => {
do {
cfa.init("#000");
neuron(cfa.identity);
} while (cfa.queue.length < 2000);
cfa.center(1.5);
cfa.render();
};
start();