Usine provides several data filter modules for smoothing, stabilizing, and shaping data flows. Unlike audio filters that process sound signals, these filters operate on data values to remove noise, smooth transitions, or add physical behavior to parameter changes.
| Module | Method | Character | Use case |
|---|---|---|---|
| Smooth | Exponential smoothing | Transforms abrupt value changes into progressive transitions, configurable time in ms | General-purpose smoothing, UI knob smoothing, data-to-audio rate conversion |
| Smooth Attack Release | Independent attack/release | Separate rise and fall times, works on arrays with per-element smoothing | Envelope following, asymmetric smoothing, array processing |
| Moving Average | Rolling window average | FIR-type smoother, averages N recent values for progressive noise reduction | Sensor data smoothing, trend extraction |
| Median Value | Sorted window median | Rejects outliers by returning the middle value of a sorted queue | Spike removal, robust noise filtering (values in 0..1) |
| Kalman | Kalman filter | Statistical prediction-based filtering, optimal for noisy measurements | Sensor fusion, noisy input stabilization (optimized for 0..1) |
| Hysteresis | Dead-zone threshold | Only passes changes larger than the hysteresis value, blocks small fluctuations | Jitter removal, threshold-based gating, noisy controller cleanup |
| Spring | Spring physics model | Output follows input with inertia, overshoot, and oscillation (stiffness, damping, mass) | Bouncy UI animations, physics-based modulation, organic transitions |
For general smoothing:
For noisy data:
For jitter and small fluctuations:
For physical / animated behavior:
version 7.0.250121
Edit All Pages