Feature Library

Ferrofluid pools & field lines

0.20 GPU · 0.19 CPU ms
A still of the Ferrofluid pools & field lines demo
Still
The live demo starts when it’s in view.Drag the cards together to fuse their pools · hover a card to raise its spikes

Liquid metal under every card: pools that fuse as cards come close, beads and a living ripple, Rosensweig spikes rising toward the pointer, iron filings and…

What it costs a frame

0.20 ms
0.19 ms
GPU
0.20 ms
CPU, median
0.19 ms
CPU, 95th pct
0.34 ms
Draw calls
6
Weight
Light

Measured 3 November 2026 on RTX 4060 Laptop, 13th Gen Intel(R) Core(TM) i9-13900H, Chrome 156, 1920×1080 at 1×. 600 frames × 3 rounds, median kept. What ran: Measured at rest with the pointer off the board: the pools, field lines and their living surface; the spikes and filings add a little while a card is hovered. Phones: tiers are being measured next.

On every machine it was measured on
MachineGPU msCPU msCPU p95
RTX 4060 Laptop · 13th Gen Intel(R) Core(TM) i9-13900Hreference · measuring tool0.200.190.34
apple · Apple A19 Proin a browser tab · 16 November 20261.310.220.44
Intel Arc 140V (Xe2-LPG) · Intel Core Ultra 7 258Vin a browser tab · 15 November 20261.700.490.88

Bars against one 16.7 ms frame. A browser-tab run is timed in a visible tab without the tool’s scripted input, so read it as close, not exact.

How steadily it runs on Intel Arc 140V (Xe2-LPG) · Intel Core Ultra 7 258V · in a browser tab

Typical frame ?
6.91 ms
1% low ?
102 fps
Varies by ?
±1.27 ms
All metricsIntel Arc 140V (Xe2-LPG) · Intel Core Ultra 7 258V · in a browser tab
p50p95p994.08 ms9.67 ms +
How long each frame took. Most frames sit in the tall bars; a long tail to the right is hitching.

Frames run back to back, each starting once the GPU has finished the last, so frame times show the work itself. They have a floor of about 4–5 ms from the browser’s own scheduling: GPU and CPU ms stay the measure of cost.

Frames

Median (p50) ?
6.91 ms
p95 ?
9.06 ms
p99 ?
9.63 ms
Slowest ?
10.80 ms
Mean ?
6.96 ms
Standard deviation ?
1.27 ms
1% low ?
102 fps
0.1% low ?
93 fps
Hitches ?
0 of 600
Over budget ?
0 of 600

Main thread

Long tasks ?
0
Long frames ?
0
Event-loop delay ?
0.1 / 1.0 / 1.1 ms

Input

Interactions ?
79
Input to paint ?
4.8 ms, worst 8.6
Longest handler ?
0.1 ms

Memory

JavaScript heap ?
74.9 → 74.2 MB
Growth ?
-0.73 MB

What it needs

Renders with
WebGPUNeeds WebGPU: current Chrome and Edge, Safari 26+, Firefox 141+ on Windows.
Your browser
Checking…Looking for WebGPU
Depends on
three.js 0.187.0No other packages

What it’s for

  1. Canvas and whiteboard apps
  2. Dashboards with linked cards
  3. Node graphs
  4. Hover states with character
  5. Product hero sections

Pitfalls, in brief

  1. Each pixel is drawn by the pool whose card is nearest (the others discard it), so fused pools are drawn once; a pool knows at most four neighbours, so a dense cluster can show seams.
  2. setPools and setRoutes compare a key of their input and only upload on change: call them every frame freely, but do not mutate the arrays in place expecting an upload.
  3. Board units are CSS pixels here: pass unit (board units per CSS pixel) and pixels (device pixels per unit) to update, or the fine detail melts or sparkles at zoom.
  4. The camera is orthographic with y down (top < bottom), which mirrors the projection: the materials are double-sided for that reason.
  5. update() returns whether it still moves: a render-on-demand app should ask for another frame only while it returns true (or while the surface is alive).
5 pitfalls in full, the hook-up guide, the module source and its tests come with the part.

Where it comes from

From SpaciousPro. Fully custom code.

Inigo Quilez, 2D distance and gradient functions: The rounded box’s distance and gradient follow his sdgBox formula (re-written in TSL).

Rosensweig instability: The spikes imitate the peaks a ferrofluid forms in a magnetic field.

sdfsmooth unioninstancingiridescencecanvasboardtslhover

Near it in the frame