Transient-Gated Emissivity: Making Pixels Scream

Moving geometry to the beat is only half the battle. To make a visualizer feel genuinely aggressive (especially for genres like Breakcore or Techno), we have to manipulate the fragment shader's lighting model using audio transients.

The Standard Lighting Model

In a raymarcher, surface color $C_{out}$ is usually a combination of base albedo $C_{base}$, Lambertian diffuse $L$, and a Fresnel rim-light $F$ to highlight edges.

$$F = (1 - \max(0, \hat{n} \cdot \hat{v}))^p$$ $$C_{out} = C_{base} \cdot L + C_{glow} \cdot F$$

The problem: If you just scale $C_{glow}$ by a smoothed audio envelope, the whole shape just gets brighter and darker. It lacks violence.

The Solution: The Overdrive Multiplier

We take a high-frequency transient metric—like Complex Spectral Difference (CSD), which measures phase-deviation in the audio signal to detect sharp, atonal hits—and use it as an overdrive multiplier squared or cubed.

Let $\tau$ be our sharp audio transient (normalized $0 \to 1$).

$$C_{out} = (C_{base} \cdot L) + C_{glow} \cdot F \cdot (1 + \lambda \cdot \tau^k)$$

By raising the transient to a power ($k = 2$ or $3$), we create a mathematical noise gate.

• When $\tau = 0.2$ (quiet hi-hats), $\tau^3 = 0.008$ (essentially invisible).
• When $\tau = 0.9$ (massive snare), $\tau^3 = 0.729$.

The Fresnel term explodes in brightness only on the absolute hardest hits, and decays instantly. Because it is multiplied against $F$, the flash only occurs on the grazing edges of the geometry, creating a sharp, localized electrical spark rather than washing out the entire screen.