A rack of media servers in a dark machine room, their status LEDs the only light.

TOOLS

Live visuals software: Resolume, Notch and TouchDesigner against the media server

None of this is drawn by hand at showtime. A small, stable set of programs does the work, and most of the craft is knowing which of them owns which job.

01

Instruments and construction kits: Resolume against TouchDesigner

seven pages: five programs, two calculators

The short answer to which software live show visuals are built in is a small one: five programs and two calculators cover almost everything a touring rig does. The toolchain divides along a line that is easy to miss: some of these programs are instruments and some are construction kits.

An instrument has strong opinions about what a show is, and in exchange it is fast to play under pressure. Resolume is the clearest example; it assumes clips, layers and a tempo, and that assumption is why an operator can work it without looking. A construction kit assumes almost nothing, which makes it capable of anything and considerably more work; TouchDesigner sits at that end, and writing code is the far edge of it.

Most rigs use both kinds. The instrument runs the show; the construction kit handles the part nothing off-the-shelf covers, and hands its output back as another source.

Live visuals software is therefore best read as a division of labour rather than a shortlist. The program that plays the show back is rarely the one that generated the material, so the question of which software live show visuals are built in has two halves: what runs at showtime, and what made the content weeks earlier.

Why the distinction decides the media server rig

A hardware control surface on a road case in an empty dark venue: a grid of dimly lit square trigger pads with one lit magenta and one amber, a row of small controls down the right-hand edge, cabling trailing away.
Playable without looking · the whole argument for an instrument

The question is never which program is better. It is which one can be operated at the moment it matters. An instrument constrains what you can express and in exchange makes the remaining moves reliable under fatigue, in the dark, with a room waiting. That trade is worth taking for anything on the critical path.

A construction kit inverts it. There is no shape it cannot be bent into, and equally no shape it arrives in. Everything (the interface, the failure behaviour, the thing that happens when a file is missing) has to be built, and whatever is not built is a hole that only appears under pressure.

The rigs that work put the kit upstream of the instrument. The custom part generates or transforms something, hands it over as a normal source, and the instrument plays it. The rigs that go wrong ask the construction kit to also be the thing an operator drives during the show.

What each piece of live visuals software is for

Each piece of live visuals software below is described by the job it owns rather than by its feature list, because the feature lists overlap and the jobs do not.

Resolume is live playback: a grid of clip banks composited across layers and sent to the show output. It exists for the case where the picture is played rather than run to timecode, which makes it the default answer to what a VJ is running.

Notch computes content during the show instead of playing content decided last week. That is the whole argument for it and the whole risk: a picture that can react to a live audio feed or a tracked object is also a picture nobody has seen before the audience does.

TouchDesigner is neither a player nor really a renderer. It is the systems tool, reached for when the problem is not “show this content” but “connect these things and make them behave”, which is most of what a large rig needs and none of what a clip player offers.

MadMapper does one job that every playback tool does badly at scale: aligning an image to a surface that is not flat, and re-aligning it every single day. Where the geometry is complicated, a dedicated mapping stage earns its place in the chain.

Cinder is the far edge — a C++ framework, reached for when the requirement is novel enough that configuring something is harder than writing it. It appears in live work rarely and for good reasons when it does.

The two calculators

Two of these pages are not reading at all; they are working tools that run in the page.

Throw distance resolves lens, image width and distance against each other. It is the most-repeated sum in projection work and the one most often done wrong on site, usually because throw ratio is treated as a property of the projector rather than of the lens fitted to it.

Pixel pitch works out the coarsest pitch that still resolves at the nearest seat, and what the resulting wall costs a media server in pixels. Over-specifying pitch is the most common and most expensive error in an LED specification: two steps finer than the room requires looks identical from every seat and costs considerably more to buy, fly and power.

What Notch and the rest do not decide

No live visuals software decides what the show looks like. The toolchain determines what is possible and what is survivable; the picture itself is settled earlier, by art direction and by the geometry previs produces. A rig assembled from the best available tools with nobody holding that seat produces something competent and generic, which is a failure the tools cannot see.

How the picture gets designed is under stage and scenic design; how it gets run is under real-time and VJ craft. The productions these programs were used on are documented in the archive, and every term used here is defined once in the glossary.

FAQ

Questions

the things people ask about this

Which tool does a live show actually run on?

Most commonly Resolume for playback, with a dedicated mapping stage after it and real-time blocks loaded as sources where reaction is needed. Very large timecoded shows may run a commercial media server instead, with the same division of labour.

Is generative content replacing playback?

No, and it should not. Playback is predictable, cheap to run and known-good before the doors open. Real-time earns its place specifically where predictability is what you do not want, a live audio feed, a tracked object, an improvising performer.

Why use a separate mapping tool?

Because calibration is a daily job with its own operator, and separating “what the picture is” from “where it lands” lets two people work at once. On simple surfaces the playback tool’s own mapping is sufficient.

Does anyone still write custom software?

Less often than a decade ago, because the standard tools absorbed most of what used to require code. It persists at the edges: unusual hardware, research-adjacent work, and techniques being developed rather than deployed.