Creative Effects

SATURIFT

Eight saturation models separated by physical mechanism rather than curve shape.

£49 £65Launch price

See the measurements

Formats: VST3 · AU · Standalone
Operating system: macOS · Universal
Version: v1.0.0

What it does

Eight circuits. Eight mechanisms. One decision.

A push-pull valve stage does not sound like a single-ended one because its topology cancels even harmonics as a matter of algebra. A transformer saturates the bass before the treble because flux is the integral of voltage. A diode clipper has no closed-form transfer function at all — its node equation is transcendental and has to be solved.

SATURIFT models those things as the physical effects they are. The differences between its eight models come out on their own, instead of having to be dialled in.

Around the models sits a complete chain: pre-EQ ahead of the drive so the low end stays defined, emphasis shelves that are exact inverses so only the harmonics survive, a Chebyshev harmonic injector that means what it says, and an auto-gain stage that measures the nonlinear section rather than calculating it.

Key features

Eight models, eight mechanisms

TRIODE, PENTODE, VARI-MU, TAPE, TRANSFORMER, CONSOLE, DIODE and GERMANIUM. Four have a memoryless curve and are antiderivative anti-aliased. The other four — tape hysteresis, transformer flux and two Newton-solved diode nodes — carry state and rely on the oversampling cascade instead.

A harmonic core that is a separate stage

Chebyshev polynomials are the only basis where the nth polynomial on a sinusoid gives the nth harmonic and nothing else. That is what lets a control called “more fifth harmonic” mean it. Even/Odd tilts the balance without ever muting either family.

Emphasis is not EQ

Low and High Emphasis boost a band into the nonlinearity and cut it by exactly the same amount coming out, so the pair's frequency response is flat and only the harmonics the boosted band generated survive.

Auto-gain measures rather than calculates

Matched mean-square followers either side of the nonlinear section derive the makeup. Deliberately slow in steady state — a fast level matcher is a compressor, and a compressor there would flatten the dynamics the saturation was chosen to shape.

Model switching is crossfaded

Both models run for 25 ms and the engine blends between them, advancing the weight per oversampled sample rather than per block. Measured across all 56 model-pair transitions, the worst inter-sample step is 0.058 against the tone's own per-sample slew of 0.014.

Oversampling order is computed, not chosen

Holding the stopband constant and letting the transition width be whatever the rate allows means the same target costs 123 taps at 44.1 kHz and 23 at 96 kHz — with latency falling from 68 samples to 17.

The interface

SATURIFT driven hard, with the HEAT needle high and the segment meters reading into the red. — SATURIFT · 1 of 2

SATURIFT with the tone stack engaged, the spectrum display drawing the filter response behind the TONE controls. — SATURIFT · 2 of 2

How it works

The signal path, as the plugin’s own documentation states it.

input trim
  → [ M/S encode ]
  → PRE EQ                       low cut, high cut, shelves, tilt
  → punch / dynamic-bias detection   (stereo linked)
  → drive
  → emphasis                     boost into the shaper, exact inverse coming out
  → LF conditioning              bass held back from the shaper, restored after
  → ⇑ OVERSAMPLE
        STAGE A   model core                    (character)
        STAGE B   density shaper                (knee tightens with drive,
                                                 ceiling sags with programme)
                  + Chebyshev injector          (Harmonics / Even-Odd)
    ⇓ OVERSAMPLE
  → LF restore → HF damping      corner falls as drive rises
  → de-emphasis → DC block
  → density                      program-dependent compression
  → auto gain                    measured across the nonlinear section
  → POST EQ                      presence, air, tilt, shelf
  → [ M/S decode ]
  → mix against the delay-matched dry
  → output trim

Audio

Audio demos for SATURIFT are not published yet.

We would rather show you nothing than show you a demo that was not recorded through this plugin.

Measured

SaturiftPluginTests and SaturiftBench · Apple Silicon · 48 kHz

Worst alias — 7 kHz at +18 dB drive into TRIODE, 16×: −100 dB
Worst alias — same case, 2×: −36 dB
CPU — TRIODE / PENTODE / VARI-MU at 4×, stereo: 2.4 %
CPU — DIODE / GERMANIUM at 4×, stereo: 4.7 %
Latency at 44.1 kHz: 68 samples
Latency at 96 kHz: 17 samples
Null at Mix = 0, every oversampling factor: exact
Worst inter-sample step across all 56 model transitions: 0.058

Figures are produced by this plugin’s own test suite and benchmark, on the hardware and sample rate named above. They are not estimates.

Controls

Drive & character

Drive: How hard the signal is pushed into the nonlinearity.
Character: Each model's own primary morph — deliberately not the same quantity in each.
Harmonics: Harmonic complexity, independent of Drive.
Even / Odd: Harmonic balance, monotonic, never muting either family.
Sag: Supply collapse under load.
Bias: Static asymmetry.
Warmth: Low-frequency and even-harmonic weighting.

Dynamics & tone

Punch: Transient-driven drive modulation, so attacks saturate harder than the sustain behind them.
Density: Program-dependent compression after the saturation.
Low / High Emphasis: Which part of the spectrum drives the distortion.

Routing & output

Mix: Delay-matched, so a parallel blend is a true blend.
Stereo Link: How much each channel's detectors see of the other.
Routing: Stereo, mid/side or dual mono.
Bypass: Wired to the host's own bypass parameter and implemented as a full dry mix, so it is bit-transparent and does not change latency.

Technical details

Formats: VST3 · AU · Standalone
Operating system: macOS 10.13 or later
Architecture: Universal — Apple Silicon and Intel
Plugin code: Crnh
Version: 1.0.0
Latency: 68 samples at 44.1 kHz, 17 at 96 kHz — reported exactly, and exact
CPU: 1.0–14.0 % of one core depending on model and oversampling factor
Factory presets: 85

Across all eight models and seven categories: vocals, low end, drums, bus, mastering, instruments and extremes. Oversampling is part of the preset and is not always high — stateful models run at 8× where the ADAA models are comfortable at 4×.