The idea here is mainly that a single html element would be enough to represent all kinds of visual content. With enough box-shadows set on a 1x1px pseudo-element, one could theoretically paint out entire images, pixel by pixel.
This is all the HTML necessary:
<div id="p"></div>
And here’s the CSS:
#p {
width: 400px;
height: 450px;
background: black;
margin: 0 auto;
margin-top: 10px;
margin-bottom: 2em;
position: relative;
display: block;
border: solid 11px black;
box-sizing: border-box;
overflow: hidden;
}
#p::after {
content: "";
width: 12.4%;
height: 10%;
background: white;
position: absolute;
top: 316px;
right: 0;
box-shadow: 0px 55px 0px #ffff00,
0px 95px 0px #ffff00;
}
#p::before {
content: "";
width: 70.6%;
height: 70.6%;
background: #ff0000;
position: absolute;
top: 0;
right: 0;
box-shadow: -280px -205px 0px #ffffff,
-279px -188px 0px #000000,
-280px 0px 0px #ffffff,
-280px 316px 0px #0000ff,
208px 315px 0px #000000,
0px 316px 0px #ffffff;
}
Of course, probably the best way to get a single-div mondrian representation
is to set a background image of mondrian.jpg
.