RigSculpt + Workshop

RigSculpt began as my own adaptation of Steve Atkinson's open-source Neural Amp Modeler plug-in (a personal build with the features I was missing) and grew from there into a complete amp-and-effects workstation. The neural amp model is the heart; everything around it: pickup voicing, drive, compressor, modulation, delay, the five reverb engines, output EQ, the Synergize stereo-matching macro, stereo routing, the library browser: is my own design added on top.

The five tabs

The interface is organized left-to-right roughly in signal order:

AMP is the boot tab. A gear icon opens Settings.

1. Run the RigSculpt installer; the VST3 lands in your system VST3 folder. Re-scan plug-ins in your DAW.
2. Insert RigSculpt on a guitar (DI) track. It boots a factory preset so you hear something immediately.
3. Open Settings (gear icon) → set your audio interface (or type its instrument-input level in dBu). This is the single most important setup step: see Input calibration below.
4. Go to NAM LIBRARY, point it at your folder of models (Folders ▾ → Add folder…), and load one onto AMP A. Add a cab on the CAB LIBRARY tab.

Everything about the amplifier itself: which model(s) play, how hard you hit them, the output EQ, pickup voicing, the tuner, and the stereo dual-rig.

Loading a model

Each amp path has a model loader. Drag a model in from the NAM LIBRARY tab, send one from the Workshop, or use the loader's file browser: and clicking the loader box itself opens the NAM Library Explorer (switchable back to the classic in-place file menu in Settings; the arrows and folder icon are unchanged either way). A small loaded-slot strip shows what's in A and B; click it to jump to that model in the library inspector.

Full-rig captures are handled properly. RigSculpt recognizes cab-inclusive captures however they're tagged (full-rig, amp_cab, amp+cab, and friends). Loading one onto a side switches that side's cab off automatically, so the capture's own cab isn't doubled; loading an amp-only capture back restores it. A manual cab toggle always wins, and the Full rigs bypass the cab automatically switch in Settings turns the whole behavior off. Across a stereo cab pair (0.97): if the other side has no amp loaded, or amp stereo is off, its cab is bypassed as well, because the one amp chain feeds both cabs and the rig's own cab must not run through a second IR; load a real amp on that side and its cab comes back.

One-button fits: RIG FIT & BUILD STEREO Workshop data

Two green buttons sit side by side at the top right of the NEURAL AMP MODEL panel. Both are one-shot compositions over the measured fits documented throughout this guide: they stage controls from your rig's Workshop measurements, report exactly what they did, and leave everything hand-tweakable.

RIG FIT stages the whole chain to the loaded amp. Click it, then play normally for a few seconds, your loudest playing included. Instantly: the Drive's Amp Match and Hold Level engage, the Compressor's Rig-Aware eases to the amp's measured squash, the Delay and Reverb are voiced to the amp (tone, ducking, mix, decay, low and high cut), and TAME's depth is prepared. The listening pass then finishes the job: the Compressor's Expert lens anchors to what you actually played and the Output lands on your target peak. A report card lists every move, line by line; click it away and tweak from there. An amp without Workshop analysis gets honest level staging only (and the card says so); effect power switches are never flipped.

BUILD STEREO (to its left) turns the loaded amp into a finished stereo pair from your own library: it suggests the best measured partner (same family, complementary brightness and character, never a near-duplicate), loads it to AMP B, spreads the pans wide, gels the two amps with Synergize so they meet in the middle, then balances and levels the pair while you play. By default it prefers a DIFFERENT amp model on the other side (other NAMs of the same amp only appear when nothing different qualifies); right-click the button to toggle that preference off for a same-amp wide-double instead. The report card names the partner and its pairing score, and lists every move it made (all of it hand-tweakable, Synergize included).

The PLAY NOW card. Every measured listening pass (RIG FIT, BUILD STEREO, Match L/R, Level) opens an unmissable prompt: play as LOUD as you actually play. Its progress bar advances only while the plugin hears you (silence doesn't count), and it shows the loudest peak heard so far. The card stays up until enough playing has been heard; click it to cancel (nothing is applied on a cancel).

Output EQ & Character

RigSculpt's own 8-band graphic EQ, plus a tone Character control for post-amp shaping: voicing the sound after the neural model without retraining or touching the model itself.

The eight bands are Sub-bassBassLow-midMidHi-midTreblePresenceAir.

Workshop data The EQ view can display the model's measured response as its reference curve, so you're shaping against what the amp actually does rather than a flat line. Requires the model to carry framing.

Curve-first rendering. The eight sliders define the curve itself: RigSculpt draws one smooth line through all eight settings and solves for the filter settings that produce that line, instead of stacking eight independent filters and hoping. Sliders land where you set them (the old law could miss by 2 dB or more), neighboring bands cooperate into one wide lift instead of two stacked humps, and width follows intent: small moves paint broad, full-throw moves focus, cuts run slightly wider than boosts. SYNERGIZE writes through these same bands, so stereo matching inherits the smoother rendering. Nothing about your presets changes on disk; the same eight values simply render more faithfully, and closer to what the sliders always claimed.

The panel draws the real curves. The Changes view traces the exact response the EQ renders, threading through your fader caps; the Measured view draws the model's Workshop-measured shape as a smooth dim ghost with your result (measured plus EQ) bright on top. Both views keep their meaning, the faders keep their gestures, and Flatten, Accentuate and Neutral behave exactly as before.

Two Settings options complete it, both off by default and per amp path in stereo. Output EQ holds your loudness rides the engaged EQ with a measured makeup gain, weighted by where the loaded amp actually puts its energy, so moving the bands changes tone at steady volume: a bass boost stops doubling as a volume boost, and the gain eases in and out so nothing clicks. Output EQ bands follow your amp Workshop data re-centers the inner bands on the loaded amp's measured structure: Mid moves to where this amp's midrange actually dips, Presence to where its top end actually peaks, the neighbors re-space smoothly, and every band label shows its working frequency ("Mid 440") with moved bands tinted. Amps without analysis, or without a clear dip or peak, keep the standard layout.

TAME Workshop data is a narrow cut parked exactly on the loaded amp's measured resonance peak, in the right-hand gutter under the level target. Many NAMs carry one dominant peak that turns fatiguing at volume; TAME reins it in without re-voicing the amp (fixed musical width, a Depth knob, per-path frequencies in stereo). The frequency is read from the file's Workshop analysis, never guessed: no measured peak means no effect, and the readout under the toggle says why.

Pickup Voicing

A physical model that re-voices your guitar's pickup toward a target (e.g. make a bright single-coil sit like a fatter humbucker, or vice-versa) so a model made with one pickup translates to your guitar.

This is the quick panel. The deep editor: including the new CUSTOM mode, where you design your own pickups from a basis plus three physical macros and save them under My Pickups: lives on the FX CHAIN page's IN section (see The IN section below).

v2 "physical" engine: families built from inductance / self-capacitance / resistance / eddy primitives, resonance derived as one reversible biquad.

Tuner

An always-on chromatic tuner lives in the AMP header. It reads your dry DI (pre-amp), so it works no matter how saturated the patch is. Defaults to A440; reference pitch and temperament are remembered.

Right-click the readout for the deeper tools: 12 tuning presets (standard, drops, opens, DADGAD and friends), per-string sweetened offsets, a strobe display mode, and auto/coarse/fine needle response. The engine reads down to 40 Hz (low enough for 7/8-string and bass) and holds a decaying note all the way down instead of dropping the reading early.

Reading the cents. The exact cents offset now stays on screen the whole time: it no longer collapses to just an IN TUNE pill when you land, so you can watch how steadily you are holding it (the number turns green in the pill when you are in tune). The needle also settles faster the further off you are and eases in as it closes on pitch. If a note-length ("period") tuner such as Reaper's ReaTune ever reads a hair different, that is expected, not an error: RigSculpt locks to the string's fundamental partial, the strobe-tuner convention, while a period tuner averages the whole waveform. On a real, slightly inharmonic string the two disagree by roughly one to three cents by design; on a clean tone they agree to a fraction of a cent.

Stereo dual-rig

RigSculpt can run two completely independent amp + cab paths at once (different model, cab, EQ, gate and architecture per side) then blend and pan them into a stereo image.

Match L/R & the target peak. One click balances the two amp paths to equal measured level AND sets the Output so your loudest playing lands on the target shown beneath the button (click the "TARGET −0.5 dB" badge to type a different one; it persists). In mono the same button reads Level and just does the leveling. Either way a PLAY NOW card opens and listens until it has heard enough real playing. Play as loud as you actually play, since the loudest peak IS the measurement.

Reading the meters. The input and output meters carry a numeric peak-hold readout in dB: it holds the loudest recent peak so a fast transient doesn't vanish before you can read it, clears itself after silence (or click to reset), and on an actual clip it turns red and latches until you clear it, so an overload can't slip past unseen. In a stereo rig the input meter shows both sides too, matching the output meter.

Mono-safe by measurement. Before it balances the levels, Match L/R runs a short deterministic probe through both amp paths offline and compares whether the two sides are in phase with each other, so the result is repeatable and doesn't depend on what you happen to be playing. If a partner NAM happens to be wired opposite-polarity, the two sides cancel when summed to mono, heard as hollow, phased delay repeats and a thin sound on any mono fold-down (a phone speaker, a mono PA). Match L/R flips that side for you and flashes B FLIPPED, so the pair sums to mono without that cancellation; loading a new amp on either side clears the flip (it belongs to the pairing, not the slot). BUILD STEREO runs this same probe, so a partner it picks is corrected automatically, and its report tells you whether that partner is a different amp or another NAM of the same one.

Automatic on a library load. You do not have to press Match L/R for a stereo rig you build from the library. When you load a partner amp into the pair from the NAM LIBRARY EXPLORER, RigSculpt runs the same level-and-phase match on its own, quietly, while you play, and shows a brief Auto-leveled A/B note when it lands. That note also says what it did about phase: Phase flipped on B if it inverted the partner to keep the pair mono-safe, or Phase in sync if it did not need to. If the deterministic phase check finishes a moment after the note appears, the note corrects itself in place. You can still press Match L/R at any time to redo it by hand.

Stereo input: left into A, right into B

RigSculpt isn't only used for guitar: people run NAM models on keys, drum buses, whole mixes and stereo outboard captures. By default the plug-in folds whatever it's fed down to mono before the amps, because that's the honest thing to do with a guitar DI. STEREO IN (the toggle next to STEREO) changes that: the left channel feeds AMP A and the right channel feeds AMP B, each side gated and processed on its own, so an already-stereo source keeps its image all the way through instead of getting collapsed and rebuilt. Pan defaults (A left, B right) already match.

The DRIVE section can create a two-signal rig even from a mono guitar: with the amp section in stereo and the drive's ROUTE set to A or B, the routed amp hears the driven signal and the other amp hears your clean pre-drive signal, hard-panned left and right like any stereo rig. STEREO IN stays what it is: a true stereo pair on the track.

• When it engages: stereo amp mode, a stereo track feeding the plug-in, and a model loaded in both A and B (the same file twice is the classic dual-mono move): or no model at all, for cab/FX-only processing. With a model in A but none in B there's only one amp, so the input honestly falls back to the mono sum and the toggle grays out.
• The IN cell tells the truth: the chain strip's IN block carries a small MONO / STEREO label showing which input mode is selected; it lights up only while the two-channel input is actually running. Right-click the IN block to flip the mode from anywhere.
• It watches for you: if RigSculpt detects a genuinely stereo signal being summed while everything else is ready for stereo input, it mentions it: once per session, with pointers to both switches. If the mono sum is what you want, just carry on.

Existing presets and sessions load exactly as they were: nothing changes unless you touch the new control. Mono guitar rigs are bit-identical.

Synergize Workshop data

A one-click macro for stereo rigs: it reads the Workshop framing baked into both amps, then shifts their EQ and their breakup point toward a shared meeting point so two different models glue into one coherent stereo tone. A tilt control biases which amp the other matches toward: center meets in the middle, toward A reshapes B toward A, toward B does the reverse. BUILD STEREO engages it for you, centered, when it builds a pair; from there it is a normal toggle.

It pairs unlike amps. As of 0.94 the match is strong enough to bring a clean amp and a high-gain amp to the same breakup together, not just a little closer. The EQ pull closes most of each band's measured gap at any tilt while each amp keeps a share of its own voice, and when both amps carry a full Workshop analysis the drive match lines up how hard each one is breaking up at real playing level. The widest clean-to-high-gain pairs still cannot fully meet at the tilt extremes, by design.

It knows capture types. Synergize declines mismatched pairs (a full rig with an amp-only capture) instead of computing a confidently wrong match, with the reason in the tooltip. The automatic phase check follows the same rule: a full rig's speaker chain shifts phase for real, so on mixed pairs the auto polarity flip doesn't fire. Two amp-only captures, or two full rigs, behave as before.

Where the controls live. The on/off toggle and the A-to-B tilt slider sit on the AMP page and, since 0.94, also in the NAM Library Explorer in the strip between the two loaded amps, so you can gel a pair without leaving the browser. The two sets mirror each other. Under Settings ▸ Workflow, "Synergize amps when loading in stereo" (on by default) re-gels the pair every time you load an amp into a stereo rig, by any route; turn it off to leave your amps as they are on load.

It is non-destructive. The starting point is frozen the moment you engage Synergize, and turning it off restores exactly that. Nudges you make to a matched band while it is on are temporary, lasting until you move the tilt (which recomputes the match) or turn it off (which returns your tone). Controls it never touches, like the Output knob, keep your edits.

Implemented as an overlay applied via framing-derived targets: it freezes your base state on engage, lifts cleanly on save, and forces off then re-gels when you load a new amp into a stereo pair, so it never bakes into your stored settings unexpectedly. When both models carry a v1.1.115 input/output curve, the drive match works from it directly, lining up compression depth near the pick-attack level (about -9.5 dBFS) and holding loudness from each amp's actual measured output change.

A pedalboard of original effects. The strip across the top shows the live signal order; click a block to jump to its controls, or drag a movable block across the NAM cell to change where it sits in the chain. A block lights up when its effect is engaged (Pickup revoicing on, Drive powered, a cab loaded, and so on), so the strip doubles as an at-a-glance map of what's actually running; a separate underline marks the section you're currently viewing.

In the chain strip the modulation pedals read by their effect category: VIBE (the Séance pedal) and CHORUS (Chimera): while their own panels keep their names. The first block, IN, is the input-conditioning section (below): signal always flows through it, so its cell is always lit. The small label under it reads MONO or STEREO: which input mode is selected (right-click the IN cell to flip it: see Stereo input on the AMP page).

The IN section: clean up the signal before the amp

IN also has its own tab in the top bar (IN, AMP, FX, NAMS, CABS): a section this useful was easy to miss inside the chain strip, so it's one click from anywhere. The IN chip on the strip still works, and the two stay in sync. Click either and the page becomes the input-conditioning chain: five islands stacked in the order the signal actually flows: HUSH → GATE → NOISE FILTER → DESCENT → PICKUP REVOICER. Each island darkens when its power is off and lights when it's on, like the pedals. Three of these exist to solve noise, and they get lumped together a lot, so here's who does what: hum gets subtracted (HUSH), hiss gets ducked under your playing (NOISE FILTER), and the gate handles the gaps. They stack.

HUSH goes after hum: the steady, pitched drone your pickups pick up from mains power, there even while you're playing. Teach it with one button: press LEARN, guitar volume up, don't play for about three seconds. It measures your rig's actual hum: the exact pitch (real hum isn't always textbook 60 Hz) and the exact level of every harmonic, then subtracts a clean synthesized copy. Your guitar signal never passes through a filter here, so no phase shift and no notched notes, and the correction physically can't get louder than the hum it measured. The measurement persists across sessions and reinstalls; re-LEARN whenever your setup changes. Depth sets how much of the learned hum is subtracted.

The NOISE FILTER goes after hiss: the broadband "shhh" that has no pitch, so it can't be subtracted. Unlike a gate, it works while your notes ring out: as your sound decays toward the noise floor, the hiss fades down with it instead of hanging on until a gate slams it shut. Two bands (above and below ~1.2 kHz) each duck only when their own content falls to the floor.

The NOISE GATE covers silence: when you're not playing, it shuts the door. The Thresh knob here is the same parameter as the AMP page's Gate knob (two views, one control), and the island adds the deeper timing:

DESCENT experimental is the real-time down-tuner: drop your whole rig by half-steps down to a full octave (8va), no retuning, formant-preserved (Voicing keeps your guitar's body/pickup character as you drop). Honest history: it has been in the plug-in since the beginning, hidden in developer settings, because the engine isn't where it should be yet: it's now unlocked by popular demand and labeled what it is. It's at its best on riffs and single-note lines; full chords are the hardest job for any real-time pitch engine, so expect some smearing there while it matures. E standard is transparent either way.

It is steadier since 0.107.0, and it is tunable. The engine under DESCENT took two real steps this release. First, a pitch-tracked splice assist steadies the lowest strings, exactly where drop tunings and extended-range guitars live: low single notes hold noticeably steadier and pick attacks land cleaner, with nothing to configure. Second, a new STABILITY knob sits beside Voicing. It trades tracking speed for steadiness on held shapes: at 0 the engine behaves exactly as before, response first; turning it up lets the engine reach further back for a seamless splice, which locks in wider held chords (a power chord with an added 9th is the classic case) at the cost of being a touch slower to follow fast changes. The reported latency never moves, so nothing shifts in your DAW. Dial it by ear: fast riffing lives happily at 0, held chords like the upper half. When the DAW renders offline with Render Mode set to MAXIMUM, the bounce is baked at full stability automatically, whatever the knob says, so the printed track always gets the steadiest version of the shift.

It holds its level. The Voicing curve costs the amp about half a decibel per semitone on real playing; a matched makeup gain rides the curve, calibrated so each tuning lands on the same gentle level trajectory a studio-grade offline shift of the same material follows. Level stays consistent from E standard down without overdriving the amp as the drop deepens, and the voicing itself is unchanged.

The PICKUP REVOICER island is the deep editor for the same Pickup Voicing described on the AMP page (the AMP tab keeps the quick panel). New here: a CUSTOM mode alongside the preset conversions: pick a basis pickup for each side (what you have / what you want) and shape it with three physical macros: WIND (how much wire on the coil), APERTURE (how wide a window of string it hears) and DAMPING. Save your designs under My Pickups: they persist across sessions and reinstalls, like presets, and either side of the conversion can wear them.

What the Revoicer actually is: not a clone of any specific pickup, but a simplified electrical model of how pickups behave. At any fixed setting the result resolves to a pre-amp filter curve: but so does a real pickup swap. The knobs change the pickup's recipe and the curve is calculated from it, so you can only land on responses a real build could make: the resonant peak, its sharpness and the output level all shift together in the physically right ratios, cable loading included. Because it sits before the amp, a hotter, darker recipe drives the model into breakup differently: it isn't tinting the tone afterward.

Drive (the "Crucible" pedal)

An overdrive that runs on an oversampled "island" for alias-free clipping. Its panel wears the name Crucible (the vessel of transformation by fire) while the chain strip keeps the plain DRIVE label. Four voices. Three original circuits: Gild (transparent boost/edge-of-breakup), Bloom (the feedback soft-clipper: it keeps giving past the knee, cleans up with your volume knob), Storm (true hard distortion: the signal is clamped to ground, not ridden): and a neural fourth, Neural, a captured pedal (below).

NEURAL: a captured pedal as the fourth voice. NEURAL sits at the front of the voice row, beside Gild, Bloom and Storm, the way CONV sits among the reverb types. Pick it and load a neural capture of a real drive, boost, or fuzz pedal in front of the amp, the same file format as your amp captures; the section takes on its own look in neural mode. An INPUT knob drives the pedal model harder or softer before its clipping (your creative control into the pedal, separate from the interface calibration), and a ROUTE selector (Both / A / B) chooses which amp hears the drive: A or B sends only that amp the driven signal and the other amp your clean tone, for a wet/dry or two-flavor stereo rig. Routing works for the analog voices too.

Running one neural model into another was built to get the details right. The pedal NAM runs at the model's own rate, with the same neural engine and oversampling policy as the amps (not on the analog voices' oversampled island), so it behaves exactly as captured. With output calibration on (the default), the pedal-to-amp handoff is calibrated from the two captures' measured levels: a capture-to-capture match, independent of your interface, with a LEVEL trim when you want to push harder or back off. The Settings input calibration anchors to the first neural model your signal reaches, which is the pedal when NAM drive is engaged. Missing data is named, never hidden: an amber note under the loader says if a pedal NAM lacks Workshop analysis or a calibration figure, and what to do about it (without the figures the pedal plays at its raw trained level instead; a pedal is never loudness-normalized, since how hard it hits the amp is its character). Pedal captures keep their own folder list, separate from your amp library: clicking the NEURAL loader opens a dedicated pedal browser (your pedal folders with similarity ordering and at-a-glance badges for input calibration, output calibration, and Workshop analysis), and the Workshop's "Pedal NAMs" library folder is sourced automatically. For large collections, the browser's GROUP button sections the list by pedal (make and model), by author, or by drive character; the character labels ("Heavy drive, mid-humped", "Essentially clean") are the Workshop's own words, read from each capture's embedded analysis, so the two apps always describe a pedal identically. As of 0.96 the pedal also wears the same MODEL ARCHITECTURE chip as the amp slots, right on the DRIVE panel: an A2-class pedal switches between Full and Lite per preset, every pedal shows its class, and the chip reads amber when a global quality policy is governing it, exactly like the amps. (Previously the pedal quietly inherited amp A's architecture choice; now it has its own, and it follows a live Performance/Studio flip immediately instead of waiting for a reload.)

A live clip-curve display draws the actual transfer curve of the current settings, so you can see the knee move as you turn the knobs.

Vibe (the "Séance" pedal)

A Uni-Vibe-style optical modulation: watery, swirling movement. An animated scope in the panel shows the sweep exactly as the DSP runs it.

Chorus (the "Chimera" pedal)

A delay-line stereo chorus. The voice mode is what most changes its character, from one subtle voice to a lush ensemble to a wobble-free pitch thickener:

Delay (the "Afterecho" pedal)

A stereo delay with independent L/R times. The mode sets how the two repeat lines are routed across the stereo field:

AFTERGLOW: the held drone (every delay mode)

The labeled section along the bottom of the AFTERECHO pedal is the delay's hold, named for the part of the echo that stays lit. It shipped on Reverie in 0.97; in 0.98 it works on every delay mode. Press FREEZE and the delay captures one period of what you just played into its own lossless hold: the drone keeps every note for as long as you leave it, while the delay keeps echoing your live playing over the top. Its volume is set by LEVEL and stays put (see below), the whole section lights while a hold is running, and you release it by pressing FREEZE again.

From a footswitch (map "Delay freeze" in the MIDI menu): tap to freeze; tap again while frozen to melt or stack a new capture; quick double tap to release; or hold the switch about a second to let the whole drone go. "Delay freeze RELEASE" and "Delay CLEAR" are also available as separate actions for boards that prefer one switch per job. A latching switch freezes only while held.

A held pad is a session instrument. Once you press FREEZE, the pad is yours to keep: its volume, its voice, and the rest of the AFTERGLOW settings stay with you as you move around, rather than changing to whatever each preset stored. Switch presets, change delay modes, even switch to a preset whose delay is off, and the pad plays on at the level you set. Only your own FREEZE release, or CLEAR, ends it. A LEVEL (or voice) you set mid-session sticks the same way; the section opens at the -8 dB default the first time.

Reverb (the "Reliquary" pedal)

One convolution mode plus five original algorithmic engines, chosen with the Type chips. Three of the engines carry a second Voicing: the same knobs driving a different character underneath (for Sanctum and Aegis it is a wholly different engine), so you can audition both and keep the one you like:

Shared controls: Mix · Decay · Size · Context · Predelay · Duck · Width · Low Cut · High Cut. On the algorithmic engines, Decay and Size reshape the space live. The third engine knob is contextual and its label follows the engine: Tone on Sanctum, Tilt/Shimmer on Aegis, Drip on Tension, Ascension on Ascent, and Regen on Solace.

Solace's own controls

Shimmer is how much of the pitched cloud you hear directly; the Interval tab beside it picks the pitch, 8ve or 8ve + 5th. Regen (the third engine knob) sends the voices back through the space, so new generations stack above the last: from a halo at low settings to an ever-rising choir as you push it. Swell (the knob beside Shimmer) shapes how gently the cloud blooms in: at 0 the octave answers immediately; turned up, it rises out of the note like a slow bow stroke. The fifth is a just fifth (a pure 3:2 above the octave) rather than the keyboard's tempered one, because a tempered fifth's small error would compound each time a generation re-enters the loop.

A reverb that knows where the note came from

All the algorithmic engines are true stereo: play a note on a hard-panned amp and the reverb's first reflections arrive from that side before the tail blooms to fill both ears, the way a real room answers. The late tail is a shared, diffuse field by design; that is what real spaces (and the classic studio units) do, and it is what makes a dual-amp rig sound like two amps in one room.

Dual (the toggle next to the type chips on the algorithmic engines) is the other philosophy, for when you want total separation instead of a shared room: two independent reverbs, the left input through its own engine and the right through another set to a slightly different room size, so the right amp's reverb never reaches the left ear, and a centered signal still spreads into a wide two-rooms image rather than collapsing to mono. It costs about twice the reverb CPU, and switching it clears the tail like an engine change. Convolution already processes each side separately (so the toggle isn't shown there), and in the PRE-amp placement, which is mono by design, it has no effect.

Fit Workshop data: in CONV mode, when the loaded impulse carries Workshop reverb metrics, a one-shot Fit button reads the space's measured decay time and low-end character and moves the knobs to a sensible starting point for that space: Mix (short rooms sit louder, long halls pull back), Predelay (scaled to the decay), and Low Cut (more when the space blooms in the lows). It's a smart starting point, not a lock. Tweak from there. With no embedded metrics it says so and changes nothing.

FIT Workshop data is the one-shot staging button next to the type chips, and it serves every engine. On Convolution it reads the loaded IR's measured tail (metrics embedded by the Workshop's Reverb Bench): a longer tail gets less mix and more predelay so the dry note stays distinct, a low-blooming tail a higher low cut. On the five RigSculpt engines it is voiced from the loaded amp instead: a dense saturated amp gets less and shorter wash with more predelay, a bright amp a darker tail ceiling, a bass-heavy rig a higher low cut. It sets the knobs, reports what it did on the status line, and leaves you to tweak.

Factory reverb impulses (CONV mode)

The bundled convolution set is a collection of original reverbs: each one rendered from RigSculpt's own reverb engines and then sculpted offline. Because a stored impulse carries no real-time CPU budget, these are layered and extended far past anything a live instance could run: dozens of stacked lines, resonant banks, dispersive stages and evolving tails captured as a single snapshot. Three families, named for the same engines whose character they push furthest. Load any of them in CONV mode; they behave like any other impulse (Trim and Width apply).

Your own reverb IRs. In CONV mode a Folders ▾ button (next to the IR loader, same idea as the NAM and CAB libraries) manages where the convolution impulses come from: add folders of your own reverb/cab .wav IRs, hide or remove them, or toggle the bundled factory set. The loader's prev/next arrows and dropdown then walk every enabled folder, so your whole IR collection is one click away. (The button only appears on CONV, since the algorithmic engines don't use an impulse.) If a reverb in one of your folders is the same recording as a factory one, the loader keeps the factory copy and lists it once, so the walk doesn't fill with duplicates.

Compressor (the "Covenant" pedal)

Four voices, each a different circuit character: Sentinel (a clean, even VCA), Ember (an optical cell: smooth, with a release that lengthens the longer it has been working), Talon (a fast FET, with bite that grows as it grabs), Oath (pedal-style squash: one fixed deep ratio for long sustain).

Simple mode (the default) is three knobs that hold at any input level:

Expert mode is the classic manual box: Threshold · Ratio · Attack · Release · Makeup · Mix (Ratio is fixed in Ember / Oath).

A live transfer-curve display shows the knee, the threshold (watch it follow your playing in Simple), and a dot riding the curve at your current level: next to the gain-reduction meter.

The detector has a placement-voiced sidechain high-pass, so a chugged low E doesn't pump the whole rig: the compressor follows the note, not the bass energy. Ember's release carries a light-cell "heat" memory; Talon/Oath's harmonic color engages only with gain reduction (bit-transparent below threshold).

An in-plug-in browser that reads the Workshop analysis baked into each model and organizes your whole collection by sound.

Sources & folders

The library is the union of multiple persistent source folders plus the bundled factory content. Open Folders ▾ to manage them:

• Add folder…: point at any folder of .nam files (scanned recursively).
• Click a source: toggles its visibility (hidden but kept in the list; one click to bring back).
• Remove a source: an explicit section drops it from the list (never deletes files).
• Factory content: a checkable toggle for the bundled models.

The source list lives in a global prefs.json outside the install dir, so it's shared across every instance and survives plug-in updates. Sources are de-duplicated by path.

Ways in: archetypes, makers, packs, favorites

Pedal captures don't clutter these views: the Explorer shows amps and full rigs, while drive/boost/fuzz captures live in the NEURAL drive voice's dedicated pedal browser (see the FX CHAIN page's Drive section).

Models are grouped into six gain-tier archetypes (see the Workshop section for what each means), shown as tiles: but the archetype path is never mandatory. Click an archetype tile to open its sub-flavor menu: a tile for every measured sub-flavor in that tier, plus an All models in this archetype tile. Pick one sub-flavor or the whole tier. (Tiers with no distinct sub-flavors jump straight to the list.) A quick-access row above the tiles offers All models, ★ Favorites and Recent (your last-loaded models, any route: library, preset, or a Workshop send). The left column groups the same library three other ways: by author, by amp maker, or by pack (the release a model shipped in, handy for demoing a whole pack in order). Clicking any entry goes straight to that group's model list; the scoped archetype tiles stay one click away behind "Back to Archetypes". Search and sort narrow things down fast.

Favorites. Hover any list row (or open a model's detail) and click the star. Starred models collect under the ★ Favorites chip. Favorites and Recent live in the global prefs.json outside the install folder, so they survive plug-in updates and reinstalls; they're keyed by filename, so a moved library keeps its stars.

Unframed models (no Workshop analysis embedded) still load and play exactly like any other: they browse by name and author, collect under the "Unframed" tile, and the status line explains what analyzing them in the Workshop (free) would unlock: archetypes, measured sorts, and Find similar. As of 0.107.0 the loaded-slot strip at the top of the page also wears the same amber "!" badge the AMP-tab loaders use when a loaded model carries no analysis, and its hover tip is scenario-aware: if a same-named framed copy exists in your scanned folders it points you at that copy, if a same-named unframed twin exists it names the embed step as the likely gap, and otherwise it lays out the full analyze, centralize, embed road. The first time each session you load an unframed model yourself, a one-time explainer dialog tells the same story in full; Cancel turns it off for good, and a Settings WORKFLOW toggle brings it back.

Scrolling. The scroll indicators on this page (the folder tree, the model list, the sub-flavor grid, and the detail panel) are real scrollbars now: click or grab the band at the right edge to jump and drag, instead of wheel-only. The Cab IR library list gained the same grabbable bar. A new read cache means leaving and reopening these pages no longer rescans a big folder from scratch: the first scan reads everything once, every later visit is near-instant. That cache also powers a freshness check: the Explorer (and the DRIVE pedal picker) quietly re-checks your folders every time it opens, so files you added, re-analyzed, or moved in the Workshop show up on their own, and if your old library folder went empty because the library moved, RigSculpt re-adopts the Workshop's current location by itself. A small Rescan button next to the model count forces the same sweep on demand, and the list keeps your place unless the set of files actually changed.

Long names. When a model, author, pack, preset or IR name is too wide for the space it sits in, hover it and the full text scrolls through, then settles back, so you never have to guess what a truncated name says. It runs everywhere a name can be clipped here (the list rows, the Amp / Author / Pack column, the inspector titles, the preset name, and the loaded A/B slots) and on the Cab IR list. It only moves while you are hovering something that is actually cut off, and it never affects audio.

Solo and hide. Every row of the Authors, Amps and Packs column carries an S and an X chip at its right edge. S solos the row: only those models exist in the Explorer until you clear it, so a session can live inside one maker or one pack. X hides the row, for the models you never reach for. With the mouse over the column the S and X keys do the same to the row under it. Soloing is not selecting: selecting a pack is where you are looking; soloing a pack is what exists, so you can solo a pack and then browse its amps, search inside it, or open Similar, Partner and More like this and get answers only from inside it. A scope line above the column says what is soloed or hidden, and clears it in one click. The scope is remembered between sessions and is never saved into a preset.

Any alphabet in a file name. Models, cab and reverb IRs, presets and folders whose names use Cyrillic, Turkish, accented or East Asian characters appear in the browsers and load; on Windows they used to be skipped or fail to open.

Big libraries open fast. The first open of the library in each session used to re-read every model file, which could take minutes on a large collection. RigSculpt now keeps what it learned in a small cache file next to your preferences and reads ahead in the background, so the Explorer opens at once and fills in as it goes. The cache is only a cache: deleting it costs one slow open and nothing else.

Keyboard browsing. Since 0.94, with the mouse over the model list the arrow keys move through it: up/down step through the list and load each model as you land on it, left/right jump to the previous or next group (author, maker, pack, or archetype, depending on how you came in), Page Up/Down and Home/End move a screen at a time or to the ends, and Enter opens the selected model's detail. Relational lists like Similar and Partner have no groups, so left/right simply do nothing there. If the keys do nothing in a given host, click once inside the browser first.

Inspecting a model

Selecting a model opens an inspector with its measured character:

• Archetype + sub-flavor with a confidence read.
• EQ shape: the eight octave bands, labelled, relative to the mids.
• A toggleable plot: Voicing under drive (how the EQ shifts as you push it), THD vs level (the breakup curve), or the I/O curve (output level vs input level, with a no-compression reference line: the gap below it is what the amp absorbs; this is the same curve the Drive's Hold Level works from).
• Find similar: ranks the rest of your library by tonal closeness to this model.

The loaded-slot strips at the top show what's currently in AMP A / B; click a slot (or its EXPLORE button) to open the loaded model in the inspector, or drag a list row onto a slot to load it.

The EXPLORER also carries its own I/O band across the top, so you can audition and balance without leaving the page: an output meter (with the same numeric peak-hold and clip-latch as the AMP page), a tuner strip, and the Match L/R button with its target-peak readout. Loading a stereo pair from here is what triggers the automatic level-and-phase match described on the AMP page.

The plug-in reads analysis only from the embedded framing: there's no database. "Find similar" is a cosine match on the Workshop's 174-dimension feature vector against a baked canonical reference. The fingerprint layout is versioned, and the plug-in only compares fingerprints that match its baked reference's layout: so files framed by an older Workshop build are skipped (never mis-compared) until you re-embed them with the current version. Find similar also compares like with like by model type: an amp-only NAM is only ranked against other amp-only NAMs (amp, preamp, and amp-head count as one group), and a full-rig NAM (amp+cab) only against other full rigs, because a speaker's response dominates the fingerprint and would make cross-type matches meaningless. The model type rides in the framing, so files framed by an older Workshop build may need a re-embed to join a group.

The same idea for cabinets: browse your impulse responses by their measured voicing and declared gear, and load straight into the cab slots.

• Folders ▾: the same multi-folder source management as the NAM library (add / hide / remove / factory toggle).
• Group by: Voicing, Folder, Cluster (unsupervised tonal grouping), or Duplicates (perceptual de-dup).
• Search: matches both filenames and the embedded analysis (so "Warehouse" finds a cab whose label says it even if the filename is abbreviated).
• Load A / B or drag onto a slot; the loaded-cab strip shows what's active.

An IR's detail shows measured character (low corner, presence peak, roll-off, tilt, octave bands) kept separate from declared labels (cab make/model, speaker, mic + position, author) so measurement is never confused with metadata.

Reads the rgsc chunk the Workshop embeds in the .wav; untagged IRs still load and play, just uncharacterized. Honesty boundary: anything under "measured" came from the audio; anything under "labels" was declared by filename/tag/manifest.

Presets

RigSculpt has its own preset system (.rigsculpt) on top of your DAW's: a complete rig snapshot (model, cab, EQ, FX) in one file. Presets you saved before the rename still load: the old .namsculpt extension is accepted everywhere presets are opened or listed. The name dropdown in the toolbar lists every preset across your sources (alphabetical, with the current one checked); the prev/next arrows step the same list; Save/Load use your own preset folder. Factory stock presets ship inside the plug-in, and a fresh instance boots the first one so you hear a full rig immediately.

Preset folders. Like the NAM and CAB libraries, presets are a union of folders. The dropdown's Manage preset folders… item opens the source manager: add your own folders, hide or remove them, or toggle the bundled factory set. Saving still writes to your preset folder, and a saved preset shadows a same-named stock one: so once you tweak and re-save a factory preset, your version is the one that shows.

Snapshots: several rigs inside one preset

New in 0.98. A preset can hold four complete rigs, A through D, each a whole setup (both amps, both cabs, the drive pedal, the reverb, and every knob, mode and toggle). They live under Snapshots in the preset menu: save the rig you have into a slot, recall any slot, and map the four to footswitches (the Snapshot A to D actions in the MIDI menu). Recall only rebuilds the parts that actually changed between the scene you are on and the one you pick, so a snapshot that keeps the same amp and swaps a pedal or an effect comes up almost at once. Snapshots save inside the preset and travel with it. A useful way to think of it: a preset is a song, and its snapshots are the handful of sounds you move between while you play it. You can also import a preset into a slot without leaving the one you are on, so a performance preset is quick to build from sounds you already have.

Preset changes carry their trails. Switching a preset or recalling a snapshot no longer cuts the sound: the amp cross-fades from the old rig into the new one, and the delay and reverb that were ringing keep ringing into the new sound instead of being chopped off, so changing sounds mid-phrase stays musical. Loads run in the background while the current rig keeps playing, and stepping through a MIDI preset bank warms the neighboring slots ahead of time.

Saving a preset into a fresh folder makes that folder your preset home (the dropdown then lists what's in it), and RigSculpt offers to bring your existing presets along. To keep the list clean, saving into a general "everything" folder (Desktop, Downloads, Documents, your user root) does not adopt it as the home: the preset still saves there, but the whole folder isn't scanned. Use a dedicated folder (e.g. Documents\RigSculpt Presets) if you want it to become your library.

Audio & calibration

Pick your interface (live-filtered list) to auto-fill the input calibration in dBu, or type it. This is the same calibration described on the AMP page.

If calibration is still off by your third session, RigSculpt explains once: a single popup on why this is the most important setup step (models were trained at a specific instrument level; every interface delivers yours at a different one). It never repeats; the small amber badge stays until calibration is on.

Oversampling

Optionally run the neural model above its native rate with an anti-alias filter, to reduce the aliasing the model's nonlinearity produces on bright, high-gain tones. Off = identical to stock. As of 0.96 the control is a detented slider labeled with the true rendering rate for your session and model: at a 48 kHz session with a 48 kHz model the stops read OFF, 96k, 192k, 384k, and the labels follow if either rate changes, because the real rates change. A model that can only run at its native rate (LSTM-family) dims the control and says so, instead of offering choices that do nothing. With the graphics card on, two more stops appear past 8x, 16x and 32x; they run on the card only and are covered in the GPU section below.

The ANTI-ALIAS options are three: Auto (the default: zero-latency minimum phase while you play, linear phase when you render), Low latency, and Linear phase. Measurement showed the real differences between the old four filters were latency and phase, not tone, so the choices now say exactly that.

Playback Mode & Render Mode

Two buttons in Settings set a global quality policy for every RigSculpt instance in the DAW at once, so you can play light and render heavy without touching a single preset. Each button flips its mode; the caption beside it tells you the current state (and briefly reads …applying while a change settles: rapid presses collapse safely into one change).

• Playback Mode: STUDIO runs every instance exactly as saved. PERFORMANCE turns oversampling off and runs A2 "slimmable" models at their Lite size everywhere, for the lowest CPU while tracking or playing live. Nothing is written to your presets or project: switch back to Studio and everything returns. (A1 models are unaffected by the model-size part.)
• Render Mode: applies only while the DAW renders offline (bounce/export). MAXIMUM jumps every instance to maximum oversampling and A2 models to Full size for the render, then drops everything straight back: no realtime cost, nothing saved. The ceiling is 8× by default; a Render OS setting beside it can raise renders to 16× or 32× (offline only, so live playback and presets never see it), and the Render Mode caption states the ceiling it will use. MATCH LIVE renders each instance at its own saved settings. PERFORMANCE is a live-only overlay: it never reaches a render, so a session tracked in PERFORMANCE still prints the saved oversampling and model size (by design: nobody wants aliasing printed because a tracking switch was left on). It's best-effort: realtime bounces in some hosts don't report as renders.

When a policy is overriding what you set, the OVERSAMPLING section header wears a small badge, and a governed A2 architecture chip reads amber: so the UI always tells you the truth about what's actually running. A manual anti-alias choice (Low latency / Linear phase) is deliberately left as you set it across these transitions; only Auto rides the live/render context, and it does so without changing your reported latency mid-render. Both modes persist across sessions.

GPU: the graphics card runs the models (Windows and Apple Silicon)

The GPU button in Settings moves the neural amp models off the processor and onto your graphics card: on Windows through Direct3D 12 (any card from any maker that runs modern Windows games qualifies) and on Apple Silicon Macs through Metal. OFF is the processor, exactly as before. GPU hands each block to the card and takes the result back a little later: about 5 ms at 1x and 2x oversampling, 8 ms at 4x and 12 ms from 8x up, the same figure whether you run 64, 128 or 256-sample buffers (the card's timing jitters more at the big settings, and the buffer is sized to cover what was measured); that delay is reported to your DAW as latency, so recorded tracks still line up, and in return the card does the heavy lifting: 8x oversampling costs it about what 1x does, and a stereo rig at the highest quality setting stops showing up on the CPU meter. The caption beside the button names the card the models are running on with the exact latency it adds, counts any block the card delivered late (a late block plays as a short fade to silence rather than stalling your DAW; a card that is repeatedly late drops the model back to the processor), or says why a model stayed on the processor.

With the GPU on, the oversampling slider grows two more steps past 8x. They run on the card only, and they are a property of this computer rather than the preset: the saved setting stays at 8x, and every instance whose slider sits at 8x plays at the rate you choose here, wherever the card runs the model. A model the card cannot take runs its plain 8x. The line beneath the GPU switch reports the card's timing at the chosen rate, so you can see the cost before you commit.

The sound is identical to the processor path to floating-point precision; every model architecture was checked against it. The setting belongs to this computer, never to a preset or a project, so a preset you share never asks for a card the next machine does not have. A few unusual architectures stay on the processor (the caption says so), and if the card ever stops responding the model falls back to the processor and carries on; nothing stops. The switch takes a moment to apply (the caption reads …applying while the models are rebuilt for the card).

The switch is for playback and applies to every RigSculpt instance in the DAW at once, like Playback Mode. Offline renders have their own policy, the Render GPU button beside it, decoupled the way Render Mode is from Playback Mode: AUTO (the default) hands a render to the card from 8x oversampling up, where it measured faster than the processor (about 1.7x at 8x, 7x at 32x), and keeps 1x and 2x on the processor, which renders those faster because the card's per-block overhead dominates there; ON uses the card for every render; OFF never does. Either way a bounce has no deadline, so the card is simply waited for and a render never contains a dropped block, and a render at 16x or 32x runs the card one block at a time (its fastest way of working at those settings) with no added latency in the bounce. A computer without a card path renders on the processor whatever the policy says.

On Apple Silicon the same GPU option runs through Metal (new in 0.98; the card path was Windows-only in 0.97). Everything above applies the same way: the commit-now, upgrade-after behavior on a preset or oversampling change, the fall back to the processor if the card is busy, and the setting belonging to the computer rather than to any preset. Intel Macs stay on the processor.

Workflow toggles

The WORKFLOW band collects the quality-of-life switches: the library-copy prompt (offered when you load a model from outside your Workshop library) gains a "don't ask again" answer, re-armable here; the unframed-model explainer (the once-per-session dialog that spells out what the amber "!" badge means and how to get the Workshop analysis) can be silenced or re-armed here too; and a quiet once-a-day update check runs when you open the editor: silent unless there's actually a new release, with an off switch right beside it.

Newer arrivals in the same band: My pickup stays put across presets keeps the PICKUP REVOICER's "your pickup" side set to your guitar across preset loads and restarts, while each preset's "voice as" target still comes along as saved (the same treatment covers the custom pickup designer's source side); Level-matched delay/reverb bypass holds your level steady when delay or reverb is switched off live, from the chain dots, a MIDI switch, or automation, instead of the volume stepping up by the mix amount (the tail still stops immediately; leave it off when you're A/B-ing exactly what an effect adds, since bypass is then no longer unity level); Loader click opens the library browser switches the loader-box click between the matching library browser and the classic in-place file menu; and Full rigs bypass the cab automatically governs the automatic cab bypass described on the AMP page. The pickup and bypass options ship off by default; the loader-click and full-rig switches ship on. The Output EQ's Output EQ holds your loudness and Output EQ bands follow your amp live here too, described under Output EQ & Character.

As of 0.97 the Highlight Workshop-aware controls switch sits in this band as well (it used to carry its own header): on, the controls that read Workshop analysis wear the Workshop marker green; off, they blend into the standard look. Appearance only; the features work either way.

Appearance: Module graphics

Each FX pedal (Crucible/Drive, Covenant/Comp, Delay, Reverb, Mod) wears a painted concept-art backdrop over its metal skin, and the amp, NAM-model, pickup and output-EQ panels carry backdrop textures of their own. The Module graphics slider in Settings (formerly "FX pedal art") sets how strongly all of that artwork shows: 100% is the full painting (the default), 0% is the bare metal look, and anything in between fades it; turn it down if the artwork makes the controls harder to read. It's purely visual (no effect on sound), and it's saved globally, so it applies to every instance and sticks across sessions.

Window size & fit

On a laptop where the editor once opened taller than the screen, RigSculpt now fits itself to the display when it opens. A size button in the toolbar and a window-size slider in Settings let you set the scale by hand, and it is remembered; the themed corner grip still drags to resize.

MIDI control: any controller can drive it hands-free

RigSculpt takes generic MIDI, so a foot controller (a MIDI pedalboard such as a Paint Audio MIDI Captain, or anything that sends Control Change, Program Change, or Note) can drive it hands-free while you play: switch presets, stomp effects on and off, nudge levels, tap the delay. It works in every DAW, because every DAW can route MIDI to an effect. As of 0.107.0 the front door is the MIDI CONTROL menu: open Settings and click MIDI controller setup… in the WORKFLOW band.

The MIDI CONTROL menu opens on the BOARD: a picture of a controller, a grid of pads with a row of knobs under it, chosen from a Layout menu (pads 2 x 4 with 4 knobs, the Alesis V-style default, or pads 4 x 4 with 8 knobs for bigger controllers). Click an empty pad or knob and it immediately listens: press that switch or move that pedal on your controller and it binds there. Click a mapped pad to fire it from the screen (test a mapping before the controller is even plugged in); right-click a pad or knob to edit everything about it. Pads hold switches and steps; knobs hold the continuous "~" levels. Positions are sticky, so your board looks like your hardware and stays put.

The LIST view (the toggle beside BOARD) shows every mapping as a row, including any not placed on the board. Your setup is a plain list of mappings. Each mapping is one trigger from your controller (a CC, a Program Change, or a Note, on one channel or all of them) firing one or more actions. Click + Add mapping and the row immediately listens: press a switch or move a pedal on your controller and it binds to whatever arrived. Then click + on the row to pick what it does. Because a mapping can hold up to six actions, one footswitch can flip Delay, Reverb, and Chorus together as a single scene move.

Switches

On/off for every block in the chain: OD, Compressor, Vibe, Chorus, Delay, Reverb, Output EQ, Cab IR, plus the whole IN chain (Hush, Noise filter, Gate A and B, Descent, Pickup revoicer), Stereo A/B, Stereo input, Delay freeze (tap to freeze, tap again to melt or stack, double tap or hold to release), Delay freeze RELEASE, Delay CLEAR, Bypass, Mute-for-tuning, Preset prev/next, and Tap tempo.

Nudge steps

One press = one step: Output, OD drive, Delay mix, Reverb mix, up or down. Good for buttons.

Continuous levels (marked "~")

For a CC expression pedal or knob: the pedal position drives the parameter directly, the whole range under your foot. Targets: Output, Input, OD drive, Delay mix, Reverb mix, Comp mix, Vibe depth, Chorus mix, Descent stability, Pickup amount, and the AFTERGLOW drone's Sheen. One pedal can ride several at once (Delay mix and Reverb mix together is the classic move). These need a CC trigger; the menu says so if the row's trigger is something else.

Click a row's trigger chip to re-learn it, switch channels (Omni or one), choose Momentary vs Latching, or clear it. Click an action chip to remove that action; the dot at the left enables or disables the whole row; the × deletes it. Mappings are saved globally, so your controller layout follows you into every project and session.

Program Change → presets. Program Change N recalls the N-th preset in your preset folder, the same order the Preset +/− footswitches step through. Point your controller's preset buttons at PC 0, 1, 2… and they become instant preset recall. (A number past the end of the list wraps around, so a switch never lands on nothing.)

The MIDI preset bank gives one footswitch one specific sound. A gigging set doesn't work in next and previous; it works in "chorus sound, now." The bank is 32 numbered slots, each holding one of your presets. Load a preset, open the preset menu, and pick Assign to MIDI preset slot…: from then on Program Change N recalls slot N (numbers you haven't assigned keep the by-position behavior above), and the MIDI CONTROL action list gains a GO TO PRESET group so a learned footswitch, button, or note can fire a specific slot on boards that don't send Program Change (the same group is on a MIDI board pad's right-click, so pads bind to slots too). Slots remember presets by name, not by position, so adding or renaming other presets never silently remaps your board mid-tour; assignments are saved globally, so your pedalboard works in every project; and MIDI-driven preset changes are always silent (the "keep a copy in your presets folder?" question is skipped: a dialog box has no place in the middle of a song).

The on-screen pedalboard beta is the older board: a numpad grid, one key per action, each with its own MIDI trigger, revealed through RigSculpt's Developer Tools. Left-click a key to fire it from the screen (handy for testing before a controller is even connected); right-click a key to edit it. It drives the same engine as the MIDI CONTROL menu, and both can be used side by side:

Action

What the key does: an effect on/off (OD, Delay, Reverb, Comp, Mod, Chorus, Noise gate, Delay freeze), Bypass, Stereo A/B, Preset prev/next, Tap tempo, Mute-for-tuning, or a continuous nudge (Output, OD drive, Reverb mix, Delay mix ±).

Trigger: Learn

Choose Learn (press a switch), then press the switch on your controller. RigSculpt binds whatever it sends, whether CC, Program Change, or Note, on the channel it arrived on. No CC numbers to look up.

Trigger: by hand

Set it directly: the type (CC / Program Change / Note), all channels (Omni) vs a single channel, and Momentary vs Latching.

Momentary vs latching matches how your switch behaves. A momentary switch sends a press then a release (127, then 0), so leave it on Momentary. A latching switch alternates (127 on one press, 0 on the next), so set it to Latching so every press counts, not every other one. Both panels show a live MIDI in readout at the top with the last message received (type, number, channel), so you can confirm your controller is reaching RigSculpt before mapping anything.

Getting MIDI to an effect is a per-DAW step: in Reaper, arm and monitor the track (or feed it from a MIDI item / the virtual keyboard); in Ableton Live, put RigSculpt on an audio track and point another track's MIDI To at it; other hosts have their own MIDI-routing control. The mapping is global, one layout across every instance and project, and saved in RigSculpt's preferences. As a beta feature it's still settling; the defaults are chosen to sit clear of the reserved sustain/pedal controllers, and the shape above is where it's headed.

The looper

New in 0.98. The looper sits at the very end of the chain, after the amp, cabs and every effect, so it records and plays the finished sound, and it keeps playing when you change presets. Lay a rhythm part down on your clean, switch to your lead, and solo over the loop; the loop holds its level straight through the change instead of ducking or cutting. It gives up to a minute of stereo, set aside when the plug-in loads. Every function is a MIDI action, mapped like any other.

REC

The classic single-switch cycle: first press records, the next closes the loop and plays it, the next overdubs a layer, the next drops back to play. Stopped, a press starts a fresh recording.

STOP / PLAY

Stops playback, and doubles as PLAY: pressed on a stopped loop it relaunches from the top. Pressed while recording, it discards the take.

UNDO

Removes overdub layers newest first, one press at a time (the last four passes are kept), down to the base recording. Pressed while you are still overdubbing, it cancels the pass you are playing.

CLEAR

Wipes the loop.

A strip along the top of the MIDI CONTROL panel shows the looper at a glance (a state indicator, the loop time, a position bar) with on-screen REC / STOP / UNDO / CLEAR buttons and a MIDI SETUP button. Every looper function is also a mappable action.

The footswitch layer. Most boards do not have four spare switches for a looper. Map the LOOPER MODE action to one switch: while it is armed, the switches you already use become the looper's transport, and letting the mode go returns them to their day jobs. A switch you dedicate to a looper action, with nothing else on that trigger, fires on its own; a switch that shares its trigger with a normal mapping only takes on the looper role while the mode is armed. One switch can even run the whole looper: put LOOPER MODE and REC on the same switch, and the first press enters looper mode, then presses drive record, play and overdub. Hold a REC switch about a second to clear the loop and leave; hold a STOP switch to leave with the loop still playing.

The pad window. Entering looper mode brings up a small draggable pedalboard: four clickable pads that mirror the transport, a lamp that shows and toggles looper mode, and the loop's state, length and playhead along the top. Each pad flashes when its mapped switch fires, so you can watch your controller land on the right function. The window follows the looper, appearing when the mode arms or a loop runs and putting itself away when you are done.

MIDI SETUP walks the mapping. The MIDI SETUP button in the transport strip goes control by control, LOOPER MODE first, then REC / PLAY / DUB, STOP, UNDO and CLEAR; at each step the next footswitch press on your controller becomes that control's assignment, and the pad being assigned is highlighted so you can lay your board out to match the screen. Click a different pad to assign that one instead; any step can be skipped, and the walkthrough cancelled at any point.

With a RigSculpt instance running, the Workshop can push the selected model or cab straight in: mono auto-routes to A; stereo prompts for A or B. This is the fastest way to audition a model you just found in the Workshop.

Pedal captures take their own route: "Load in connected RigSculpt VST" on a pedal lands it in the DRIVE section's NEURAL slot and engages it, with no A-or-B prompt. The plugin reads the capture's embedded gear type, so a pedal never ends up in an amp slot.

See the Workshop → "Send to RigSculpt VST" section for the other end of this.