A wavetable synth you can see.
Mesa turns living visuals into wavetables — then paints every voice back to you as light. Feed it a shader, sculpt the frame, and watch the sound move in real time.
Feed it a shader.
Hear the light.
No other wavetable synth works like this. Drop in a .vkograph graph or a raw .metal VKO1 shader and Mesa renders it offscreen, frame by frame — reading each frame's luminance as a single-cycle waveform, then band-limiting it into an alias-free wavetable. The picture becomes the sound source.
The whole graph survives the trip.
Every modifier node you stack in the VKO1 editor is traced out of the graph and reapplied on import — so the effect that bends the picture also bends the wavetable. Different effects, genuinely different sounds.
Open the VKO1 graph editor ↗Grain + envelope + waveshaping — tanh, S-curve, edge or invert.
9-tap Gaussian blur — a low-pass that softens the waveform.
Re-samples each channel at an offset — comb-filter harmonics.
Block displacement — gritty inter-frame smear and texture.
Blend modes — multiply, screen, difference, greyscale.
Color is an oscillator.
The palette isn't skin-deep. Mesa reads the picture's luminance as the waveform, so swapping palette reshapes the brightness curve — and the timbre with it. A 3D terrain renders every voice straight from the audio thread, lock-free, no latency. All 33 VKO1 palettes on board.
Tune how it sees.
Mesa listens to light and motion. The Vision panel hands you the dials — pick the scanline, weight the colour channels, bend the brightness curve, choose how time crawls across the frames. One shader, a thousand readings.
Which horizontal slice of the image becomes the waveform — drag from top to bottom of the picture.
Read RGB, a single channel, the R−B chroma axis, or equal weight — each colour route a different timbre.
Brightness transfer: linear, gamma, S-curve, wavefold, full-wave rectify or invert.
How frames sweep the shader clock: linear, exponential, sinusoidal, reverse or bounce.
Sweep the shader's energy across all 64 frames for a living, morphing wavetable.
A complete instrument, top to bottom.
Zero-delay SVF
Per-voice state-variable filter with Cutoff, Resonance and Drive. Clean, analog-flavoured, self-oscillating.
Morph is first-class
Two LFOs, velocity and mod-wheel routed through a typed matrix — wavetable morph is a primary target. Plus a free-assign slot: pick any source and route it to morph, cutoff, resonance, level or pitch.
Eight stages, in series
A full post-mix rack: tape → flutter → waveshaper → chorus → delay → reverb → lo-fi → tape-stop. Allocation- and lock-free on the audio thread.
Stacked & detuned
An oscillator stack with unison voicing and detune, morphing across the frame axis for wide, moving timbres.
Shape every note
A full ADSR amp envelope, portamento glide between notes, and a pitch envelope that drops or rises into each hit — punchy plucks, sliding leads, percussive stabs.
Browse by category
A packed factory bank sorted into collapsible folders — pads, bass, leads, plucks & bells, keys, drones, FX textures. Save your own into any category, search across the lot.
FFT mip banks
Every frame is FFT'd at build time into a mip bank; playback picks the mip that keeps the top harmonic under Nyquist.
Run it to tape
Saturation with HF roll-off, wow & flutter pitch drift, and bit-crushing lo-fi — plus a one-shot tape-stop that decelerates the whole mix to a halt. MIDI-learnable.
Play your own audio
A sampler voice mode: import any audio, Mesa auto-detects the fundamental and pitch-tracks it across the keys — hot-swapped on the audio thread, just like a wavetable.
Two engines, blended
Layer two patches and blend them to taste, with a built-in arpeggiator to set the whole thing in motion.
Map anything
Right-click any knob, chip or button to bind a CC — every parameter, the mod matrix, tape-stop and all five Vision dials included. Bindings persist across sessions.
In time, on the network
Tempo and phase lock with any Link-enabled app or hardware on the same Wi-Fi — built on the official LinkKit, App-Store-safe.
Stack effects. Hear the difference.
Build a chain of visual modifiers — wave warp, edge detect, perspective, affine, blur and more — with per-effect wet/dry and parameter sliders. Each effect bakes into the wavetable. Active effects show as chips live on the waveform display.
Effects that reshape the sound
filmLook, bloom, rgbSplit, datamosh and mask travel straight from the VKO1 graph into the wavetable — the look and the timbre move together.
Your AI co-pilot.
Now playing.
Connect Claude Code — or any MCP-compatible agent — and Mesa becomes an instrument it can play. It writes and compiles Metal shaders against real compiler errors, sculpts the sound parameter by parameter, rolls entirely new patches, and plays the keyboard to audition what it built — a local server, off by default, reachable only from your own Mac.
- Real compile feedback — your agent writes Metal shaders, compiles them on Mesa's own GPU pipeline, and reads back the actual compiler error. It iterates until the shader works — no round-trip through Xcode.
- Roll a random patch — the same dice-roll as the UI's randomizer: an unused shader plus a whole sound built around it, one call away. A fast starting point instead of building from nothing.
- Full parameter control — every knob, the mod matrix, Vision, the FX chain, the arpeggiator, even global volume — settable by friendly name, with real ranges from a single catalog call.
- Plays the keyboard — triggers and releases notes, one-shot auditions, live pitch bend, a looping step sequence at a real tempo — so it can actually hear what it built.
- Private by design — binds to 127.0.0.1 only. Off by default, one toggle away, never reachable from your network or the internet.
Mac only · Off by default · 25 tools · localhost only
Built native.
VKO Interconnected Universe
Every app in the family speaks the same language — .vkograph graphs and .metal shaders. Patch a graph once, then play it as visuals, hear it as a wavetable synth, or cut sixteen drums out of it. One picture. Three instruments.
Where it all started. Pad-based live visuals, AutoMix and 53 effects — plus the node editor that writes the .vkograph files the whole family reads.
A synth you can see. Mesa renders your shader offscreen and reads each frame's luminance as a single-cycle waveform — then paints every voice back as light.
Sixteen wavetable voices, each cut from its own scan row through a live shader. No samples, no models — swap the image and the whole kit re-cuts at once.
The full node graph on WebGL — patch generators, stack modifiers, feed it live audio. Nothing to install. Export a .vko graph and open it anywhere in the universe.