How to Visualize the Fourier Transform | QuantumSketch
Visualize the Fourier transform by wrapping a signal around a circle and watching its center of mass spike at the signal's frequencies. Here's the animation.
Visualize the Fourier transform by wrapping a signal around a circle and tracking its center of mass: when the winding frequency matches a frequency in the signal, the center of mass spikes. Sweeping the winding frequency traces out the transform.
The "winding machine" intuition
Made famous by 3Blue1Brown:
- Take a signal in time — say a 3 Hz cosine.
- Wind it around the origin at a chosen winding frequency.
- Watch the center of mass. For most winding frequencies, the wound graph is balanced → center of mass ≈ origin.
- Hit a match. When winding frequency = 3 Hz, the loops align on one side → center of mass swings out.
- Sweep + plot the center-of-mass distance vs winding frequency → spikes at each component frequency.
That plot is the Fourier transform.
Why it reveals hidden frequencies
| Signal | What the transform shows | |---|---| | Single 3 Hz wave | One spike at 3 Hz | | 3 Hz + 5 Hz sum | Two spikes | | Noisy signal | Spikes at real tones, low elsewhere |
This is exactly how audio equalizers and MP3 compression "see" sound.
Manim building blocks
A ParametricFunction for the winding curve (radius = signal value, angle = winding freq × t), a Dot for the center of mass, and a second Axes for the frequency plot. A ValueTracker sweeps the winding frequency.
The prompt
"Wrap a 3 Hz + 5 Hz signal around a circle, sweep the winding frequency 0→8 Hz, and track the center of mass spiking at 3 and 5 Hz; plot the result."
→ Render it at quantumsketch.app. Related: Animate the Derivative, Animate Sine & Cosine.
Written by Shihab Shahriar Antor · Shahriar Labs
FAQ
Q.What is the most intuitive way to picture the Fourier transform?
Picture wrapping the signal around a circle. Take a signal over time and plot it by winding it around the origin at some winding frequency. For most winding frequencies the wound-up graph is balanced around the origin, so its center of mass sits near zero. But when the winding frequency matches a frequency actually present in the signal, the graph lines up so that its center of mass swings far from the origin — producing a spike. Sweeping the winding frequency and tracking the center of mass traces out the Fourier transform, revealing which frequencies the signal contains.
Q.Can I animate the Fourier transform without DSP coding?
Yes. Describe it to an AI animation tool: 'Wrap a sum of two sine waves around a circle, sweep the winding frequency, and track the center of mass spiking when it matches each component frequency.' QuantumSketch produces a narrated Manim animation of exactly this winding-machine view popularized by 3Blue1Brown. Manim handles the parametric winding curve, the moving center-of-mass dot, and the resulting frequency plot — you just steer it with the prompt and pick which frequencies to include.