Macro Fracture

s3g Macro Fracture gathers ten NIM-derived disruption and voice processors into one Mono, 8-channel, or 24-channel CLAP. The selected model changes the meaning of three contextual controls while REACT, MEMORY, mix, and the multichannel relationship system stay consistent.

The processor menu covers contact-like gating, sample and bit reduction, short-buffer rearrangement, two-input logic, amplitude voids, formant coloration, robot modulation, and three octave treatments. Model changes crossfade over 20 ms, and every signal path is DC-blocked and bounded.

Macro Fracture 8ch with a NIM-derived processor, contextual macro labels, coordinated lane color and bias, and bounded recurrence.

Workflow

  1. Insert the Mono, 8-channel, or 24-channel build on a matching REAPER track, configure the input pins, and begin at a conservative monitoring level.
  2. Route the result to the main speakers, a direct panner, or a stereo or quad fold-down, then begin playback. Another effect may precede that destination when its build matches the current bus width.
  3. Choose a PROCESSOR, set its three contextual controls, and use REACT to make transients push the treatment.
  4. Add MEMORY only when the processor should recur or smear into itself, then set MIX and OUT.
  5. On the multichannel builds, shape color and bias relationships across the lanes before automating the shared voice.

Versions and Routing

The plugin preserves lane identity and does not mix or ambisonically encode the bus. In the multichannel LOGIC model, the last lane wraps around to the first, so the comparison forms a closed ring.

Shared Controls

ControlMeaning
PROCESSORSelects one of the ten fracture models. Changes use a 20 ms crossfade.
INPUTApplies gain before activity analysis and the selected model.
Context controlsThe next three labels change with the selected model. Their host parameter names remain Amount, Color, and Bias so automation and saved state stay stable.
REACTLets signal energy and transients increase processing amount and briefly brighten or accelerate the model.
MEMORYFeeds a bounded version of the model output back into its input. At zero, the shared NIM kernel receives the unrecirculated signal.
MIX / OUTSets the processed blend and final level.
PANICClears processor histories, short buffers, and recurrence without changing parameter values.

Processors

ProcessorContext controlsCharacter
RELAYTHRESH / CHATTER / ASYMEnvelope-driven contact gate with hysteresis and controllable chatter.
CRUSHBITS / RATE / DITHERBit quantization and sample holding with bias-controlled deterministic dither.
SPLICEDEPTH / LENGTH / DIRECTIONShort-buffer repeats that alternate or force playback direction.
LOGICDEPTH / THRESH / BALANCEXOR-like comparison against the mono carrier or next multichannel lane.
VOIDDEPTH / RECOVER / SKEWEnvelope-following removal that opens and closes around asymmetric thresholds.
THROATMIX / VOWEL / SHIFTTwo resonant formant regions for vocal and constricted timbres.
ROBOTDEPTH / CARRIER / CHARACTERSine-to-square carrier modulation with driven output shaping.
OCT DOWNMIX / FILTER / TRACKZero-crossing octave division with envelope tracking and smoothing.
OCT UPMIX / TONE / SHAPERectification-based octave doubling with envelope removal and filtering.
OCT STACKMIX / TONE / BALANCEBlend of octave-down division and octave-up rectification.

These are the same shared kernels used by the corresponding NIM insert types. Macro Fracture adds oversampled stepping, click-free model selection, reactive control, recurrence, wet/dry mixing, and lane relationships around them.

Multichannel Relationships

The relationship preview uses a bright marker for contextual Color and a darker marker for Bias. Relationship controls are omitted from the Mono build.

Memory and Safety

MEMORY uses a sub-unity, saturated recurrence path rather than exposing an unrestricted feedback gain. The processor also clamps internal excursions, removes output DC, rejects non-finite values, and applies a final soft bound. The activity panel shows how strongly the result differs from the driven input.