Live visuals drawn from the notes, not from the audio

Audio, and nothing else. A finished mix arrives as one stream of numbers, so the software runs a frequency analysis on it and drives shapes from the bands: low energy moves this, high energy moves that, and somewhere a beat detector is making its best guess about where the pulse is.

That is not a criticism of the software. It is the whole of the information a mix contains. A spectrum cannot tell you which note was played, what key the piece is in, whether the chord being held is settled or tense, or whether the last thing that happened was a new note or somebody letting go of an old one. Everything a visualiser knows about your music, it has had to reconstruct from the sound of it.

Sonari makes the sound, so the picture is drawn from the same events rather than from the result. The Stage is given the music, not a recording of it. Every note as it arrives, and its own fade after you let go of it. How consonant or how tense the thing you are holding is. The key, as a colour that travels round the circle of fifths over about two bars rather than snapping. And how densely you are playing.

Colour is by pitch, and each palette is a stated theory of which colour a note is: Scriabin’s 1910 colour keyboard, Newton’s spectrum, a chakra set built for raga. Whichever visual you are running, the same note is the same colour, so a key change is a change in the room rather than a change in one effect.

It knows the scale, too. The kaleidoscope in the finish can take its fold count from the symmetry of the scale you chose, which means a five note scale and a seven note scale fold the picture differently without anybody setting anything. If that sounds like a detail, the scale library is the deepest thing in the product and the picture is downstream of it.

Its own window, on this Mac or on any display macOS can see: a projector on a cable, an iPad over Sidecar, an AirPlay display set up as a second screen rather than as a mirror. Sonari opens or moves the window there. It can also go to a paired iPhone, which is sent the performance rather than the video and draws it locally.

Full screen it carries nothing. The pointer disappears after three seconds and the window controls stay hidden until you put the pointer in the top left corner, so what the room sees is the work. B blacks it out and leaves the engine running, so the picture comes back instantly rather than rebuilding itself in front of an audience. Escape leaves full screen first and closes second, which is the exit you can find without looking for it.

1,000+ visuals and 3,900+ looks through them. Every one is in the application already: nothing downloads, nothing unlocks, and there is no shop.

A look opens with no controls on it at all, because most of the time the recipe is the point. Ask for more and you get six that exist on every visual in the library: colour, how hard your movement pushes the picture around, brightness, glow, trails, and that visual’s own first control. Ask again and you get everything the visual declares. The Visual Lab gives the whole window to one look, with every control it has beside a live still, a randomise, and one undo that puts the whole thing back. Nothing is exclusive to it.

You can also describe a look in words and have it built. What comes back is assembled out of the visuals and palettes that already ship rather than invented, and it lands as an ordinary look you can then take apart.

They are not the same kind of thing, and pretending otherwise would waste your afternoon. TouchDesigner and Resolume are environments you build in: a graph of operators, your own media, layers, a timeline, and outputs into whatever else is in the rig. Sonari is a fixed library of visuals with deep controls inside each one. Four things follow from that, and you should know all four before you try it.

Nothing goes in. It plays no clips, holds no images and has no timeline to lay anything on. What you see is generated, every frame, from what is being played.

Nothing comes out but a screen. There is no video feed to hand another application, so Sonari cannot be a layer inside Resolume or a source in something expecting one. The way out is a display, which means anything that can capture a screen can have it, and anything that expects a feed cannot.

It draws what Sonari plays, and only that. Your hands, a paired iPhone, a loop you built, the band underneath you. A MIDI keyboard plugged into the Mac can work Sonari’s controls, but its notes do not reach the picture, so this is not a general visualiser for the sequence already in your DAW.

It keeps its own takes. Sonari records the Stage window while it is open, because that is the thing an audience saw. With sound in the recording it captures the whole screen instead, because macOS will not let a capture of a single window carry audio from anything else.

And it is a Mac application, which for anybody running Resolume on a PC is where this ends.

One person, playing, in a room that can see it. The visuals are not a layer you are operating while somebody else performs; they are what the performance looks like, and the only instrument driving them is the one your hands are already on. There is nothing to hold, which is what makes the performer available to be inside the picture in the first place.

If what you want is a rig you designed, build it in TouchDesigner and keep it. If what you want is for the wall to show what your hands are doing while you play, that is a different job, and this is the tool that has the notes to do it with. It will also sit beside the studio you already have, which is usually where this ends up being tried first.