Hover — magnetic, on a tether
Variant: Magnetic tether
A round button that leans toward the pointer before the pointer reaches it, and the whole design problem is the leash. The pull is a third of the pointer’s offset from centre, capped at eight pixels: uncapped, a target that tracks the hand that closely stops being a target and becomes something you chase. It trails the hand by about 90ms on ease-out, so it reads as answering rather than predicting, and the glyph rides half again as far as the shell on the same beats, which is the cheapest depth cue there is — the front of the button leans further than the back. Two thresholds, two signals: being inside the pull zone (drawn here as a dashed ring, invisible in production) earns the lean; being on the face earns the darker fill, so a pointer that merely passes close never claims it was over. Release is a plain 280ms ease-out home. A spring is the obvious reach here and the wrong one: on an eight-pixel lean its overshoot is a single pixel nobody sees, and the cap is the physics that actually matters. Reach for it on one or two primary actions per screen, never on a toolbar; ten magnetic buttons in a row is a field of things flinching at you.
The source
A complete HTML document. Paste it into an empty .html file and it plays — no stylesheet, script, font, or image to fetch.
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<title>Hover — magnetic, on a tether</title>
<style>
:root {
/* Neutrals — fixed, never remixed. */
--ivory: #faf9f5;
--paper: #ffffff;
--oat: #e3dacc;
--slate: #141413;
/* Knobs — the remix API. A knob exists iff declared here and used below. */
--hue: 38.8;
--lift: 0;
--chroma: 1;
--speed: 1;
/* Accents — derived from the knobs with the family's offsets baked in,
so one hue roll moves them together. Defaults reproduce the original hex. */
--clay: oklch(calc(0.6724 + var(--lift)) calc(0.1308 * var(--chroma)) var(--hue));
--clay-d: oklch(calc(0.5797 + var(--lift)) calc(0.127 * var(--chroma)) var(--hue));
--olive: oklch(
calc(0.6118 + var(--lift)) calc(0.0713 * var(--chroma)) calc(var(--hue) + 88.3)
);
--rust: oklch(
calc(0.5408 + var(--lift)) calc(0.1357 * var(--chroma)) calc(var(--hue) - 10.3)
);
/* Clock — 3.6s, so 1% is 36ms. The pointer crosses into the pull zone
at 20.8%, onto the face at 29.9%, off it at 58.5%, out of the zone at
61.6%, and the button is home 280ms later at 69.4%. */
--cycle: 3.6s;
--t: calc(var(--cycle) / var(--speed));
--_w: min(260px, 86vw);
--ease-out: cubic-bezier(0.23, 1, 0.32, 1);
/* A hand is a sine, not a UI curve: it speeds up and slows down evenly.
Every crossing time above was solved against this curve. */
--_hand: cubic-bezier(0.45, 0, 0.55, 1);
}
* {
box-sizing: border-box;
margin: 0;
padding: 0;
}
body {
min-height: 100vh;
display: grid;
place-items: center;
background: var(--ivory);
font-family:
system-ui,
-apple-system,
'Segoe UI',
Roboto,
sans-serif;
color: var(--slate);
}
.stage {
position: relative;
width: var(--_w);
height: 170px;
}
/* The pull zone, drawn. In production it is invisible and twice the
button's diameter; here it shows only while the pointer is inside it,
because otherwise the button appears to move before anything touched it. */
.zone {
position: absolute;
left: calc(50% - 56px);
top: 29px;
width: 112px;
height: 112px;
border-radius: 50%;
border: 1px dashed color-mix(in srgb, var(--slate) 22%, transparent);
opacity: 0;
animation: zone var(--t) var(--ease-out) infinite;
}
.mag {
position: absolute;
left: calc(50% - 28px);
top: 57px;
width: 56px;
height: 56px;
animation: pull var(--t) infinite;
}
.face {
display: grid;
place-items: center;
width: 100%;
height: 100%;
border-radius: 50%;
background: var(--clay);
box-shadow: 0 4px 12px color-mix(in srgb, var(--slate) 18%, transparent);
animation: tint var(--t) ease infinite;
}
/* The glyph rides half again as far as the shell on the same beats, so
the button reads as having depth — the front of it leans further than
the back. Same keyframe times, so the two can never disagree. */
.glyph {
width: 22px;
height: 22px;
fill: none;
stroke: var(--ivory);
stroke-width: 2;
stroke-linecap: round;
stroke-linejoin: round;
animation: lean var(--t) infinite;
}
.ghost {
position: absolute;
left: 0;
top: 0;
width: 14px;
height: 18px;
opacity: 0;
transform: translate(calc(var(--_w) * 0.12), 150px);
animation:
travel var(--t) var(--_hand) infinite,
presence var(--t) var(--ease-out) infinite;
}
.arrow {
display: block;
width: 100%;
height: 100%;
fill: var(--slate);
stroke: var(--ivory);
stroke-width: 1.4;
stroke-linejoin: round;
}
@keyframes travel {
0%,
9.8% {
transform: translate(calc(var(--_w) * 0.12), 150px);
}
26.9% {
transform: translate(calc(var(--_w) * 0.5 - 26px), 104px);
}
44% {
transform: translate(calc(var(--_w) * 0.5 + 22px), 76px);
}
56.2% {
transform: translate(calc(var(--_w) * 0.5 + 6px), 64px);
}
68.4%,
80.7% {
transform: translate(calc(var(--_w) * 0.86), 26px);
}
81.9%,
100% {
transform: translate(calc(var(--_w) * 0.12), 150px);
}
}
@keyframes presence {
0% {
opacity: 0;
}
4.9%,
75.8% {
opacity: 1;
}
80.7%,
100% {
opacity: 0;
}
}
/* A third of the pointer's offset from centre, capped at eight pixels —
the tether. Uncapped, a button that tracks the hand that closely stops
being a target and becomes something you chase. It trails the pointer
by ~90ms on ease-out, so it is answering the hand rather than predicting it.
Release is a plain 280ms ease-out home: a spring here overshoots by a
pixel nobody sees, and the cap is the physics that matters.
Curves are literal in mixed-curve keyframes — a var() there is ignored. */
@keyframes pull {
0%,
20.8% {
transform: translate(0, 0);
animation-timing-function: cubic-bezier(0.23, 1, 0.32, 1);
}
29.3% {
transform: translate(-6.5px, 4.5px);
animation-timing-function: cubic-bezier(0.45, 0, 0.55, 1);
}
46.4% {
transform: translate(6.5px, -2.5px);
animation-timing-function: cubic-bezier(0.45, 0, 0.55, 1);
}
58.7% {
transform: translate(2.5px, -6.5px);
animation-timing-function: cubic-bezier(0.45, 0, 0.55, 1);
}
61.6% {
transform: translate(4px, -7px);
animation-timing-function: cubic-bezier(0.23, 1, 0.32, 1);
}
69.4%,
100% {
transform: translate(0, 0);
}
}
@keyframes lean {
0%,
20.8% {
transform: translate(0, 0);
animation-timing-function: cubic-bezier(0.23, 1, 0.32, 1);
}
29.3% {
transform: translate(-3.25px, 2.25px);
animation-timing-function: cubic-bezier(0.45, 0, 0.55, 1);
}
46.4% {
transform: translate(3.25px, -1.25px);
animation-timing-function: cubic-bezier(0.45, 0, 0.55, 1);
}
58.7% {
transform: translate(1.25px, -3.25px);
animation-timing-function: cubic-bezier(0.45, 0, 0.55, 1);
}
61.6% {
transform: translate(2.0px, -3.5px);
animation-timing-function: cubic-bezier(0.23, 1, 0.32, 1);
}
69.4%,
100% {
transform: translate(0, 0);
}
}
/* The colour belongs to the face, not the zone: being near the button
earns the lean, being on it earns the hover state. Two thresholds,
two signals, so a pointer passing close by never claims it was over. */
@keyframes tint {
0%,
29.9% {
background: var(--clay);
animation-timing-function: ease;
}
34.2%,
58.5% {
background: var(--clay-d);
animation-timing-function: ease;
}
62.3%,
100% {
background: var(--clay);
}
}
@keyframes zone {
0%,
20.8% {
opacity: 0;
}
25.7%,
61.6% {
opacity: 1;
}
66%,
100% {
opacity: 0;
}
}
@media (prefers-reduced-motion: reduce) {
/* Nothing leans and the pointer parks on the face. What survives is the
state, not the physics: the zone still lights and the face still
darkens on the same beats, a little over half as often. */
.mag,
.glyph {
animation: none;
}
.ghost {
transform: translate(calc(var(--_w) * 0.5 + 6px), 80px);
animation: parked calc(var(--t) * 1.6) var(--ease-out) infinite;
}
/* The parked pointer is only shown while the face is tinted, so the
hover state never sits there without its cause, or outlives it. */
@keyframes parked {
0%,
25.1% {
opacity: 0;
}
29.9%,
58.5% {
opacity: 1;
}
62.3%,
100% {
opacity: 0;
}
}
.face {
animation-duration: calc(var(--t) * 1.6);
}
.zone {
animation-duration: calc(var(--t) * 1.6);
}
}
</style>
</head>
<body>
<div
class="stage"
role="img"
aria-label="A pointer drifts toward a round button; once inside a pull zone twice the button's size, the button leans a few pixels toward it and darkens when the pointer is on it; when the pointer leaves, the button springs back to its seat with one small overshoot"
>
<div class="zone"></div>
<div class="mag">
<div class="face">
<svg class="glyph" viewBox="0 0 24 24" aria-hidden="true">
<path d="M5 12h14M13 6l6 6-6 6" />
</svg>
</div>
</div>
<span class="ghost">
<svg class="arrow" viewBox="0 0 14 18">
<path d="M1 1 L1 14.8 L4.5 11.6 L6.7 16.6 L9.3 15.4 L7.1 10.5 L11.8 10.3 Z" />
</svg>
</span>
</div>
</body>
</html>
The variant family
Same idea, one axis moved. Compare them side by side before you commit.