Beyond DMX
The state lives in the fixture.
State-Resident Lighting Control (SRLC): a research project by ZentrumNeueLichtkultur (ZNL) on the architecture of lighting control.
What it is about
A lighting console tells every fixture up to 44 times a second what it should be doing right now. Even when nothing changes. That is how DMX512 has worked since 1986, and today’s network protocols, Art-Net and sACN, essentially carry the same thing: continuous values. The ZNL is investigating whether there is a second way. To that end we are launching the research project “State-Resident Lighting Control”, SRLC for short.
The guiding question
Why do we keep transmitting what a fixture should do, when we might be able to tell it once what to carry out on its own, and when?
“State-resident” means that the state, meaning the lighting look, target value and timing, lives where it is executed. The console distributes it once, triggers it and receives feedback.
The hypothesis
The ZNL puts this hypothesis to the test: the continuous transmission of control values can be complemented, or partly replaced, by an architecture in which fixtures store their sequences locally, calculate transitions themselves, synchronise via a shared clock and report their state back. Whether this works better in practice than what exists today is not proven. That is exactly what is to be investigated.
The idea is not new
From 1987 the Vari-Lite Series 200 system worked on exactly this principle. Every fixture had its own processor and stored its own cue data. The Artisan console called up the cue number with a single shared command; the units executed it and reported back. The patent is titled “Computer controlled lighting system with distributed processing”.
The open, simple DMX512 won the day. Why that happened, and what of it still holds today, is one of the research questions. Our working assumption: the principle did not fail because of the idea, but because it was a closed system from a single manufacturer.
What we are not claiming
We are not claiming that DMX has had its day. Nor are we claiming that bandwidth is running out: a gigabit network carries the data of a thousand fixtures with ease. The open question lies elsewhere. Where does the state of a show live? What happens when the connection drops? How does the console know what the rig is actually doing? Decentralised systems are not automatically faster, safer or more economical. They trade familiar problems for different ones.
Three models compared
The project develops at least three architectures and assesses them against the same criteria.
- Model AFixture Intelligence. Every fixture stores and calculates for itself. The console sends events and points in time.
- Model BDistributed Edge Control. Nodes in the network manage groups of conventional fixtures and run the sequences for them.
- Model CSemantic Lighting Network. The network transmits tasks and target states. The devices translate them into movement, colour and time.
A hybrid fourth model, in which the data stream remains for live busking, is likely.
AI: possible, not necessary
SRLC works without artificial intelligence. Where AI is examined, for instance in fault diagnosis or in adapting a programme to other fixture types, it is kept separate from deterministic real-time control. It does not decide any safety-relevant movements. Creative control stays with the lighting designer.
Version and authorship
Once show data no longer sits only in the console but in a hundred devices, new questions arise: which version is the valid one? Who created it? Has it been altered along the way? Today a showfile is a file on a console, in a place you can point to. In a distributed architecture it breaks up into many parts. How authorship and the integrity of show data can remain verifiable across every device is therefore part of the research question. A control protocol alone does not answer it.
Why now
Today’s fixtures contain processors that would have filled an entire console in 1987. Time synchronisation over Ethernet is standardised. Devices describe their capabilities in open formats. And manufacturers are reworking their firmware and network technology anyway to meet new European security requirements. The building blocks are there. Whether they fit together is the question.
Get involved
The ZNL is looking for hands-on experience: from operators, system technicians, developers and anyone who has worked with Series 200.
Write to us: srlc@zentrumneuelichtkultur.de
Read on
A separate page, with a simulation you can try out, shows how such a control system could work.
How it works
How it could work
Distribute, trigger, report back: the draft with a network simulation, failure cases and a comparison in numbers.
Read how it worksOverview
All articles
Initiatives, position papers, reviews, white papers and essays by the ZNL in one place.
To the overviewSources
- ANSI E1.11, “USITT DMX512-A” (ESTA). Timing and structure of the DMX packet; this gives the repetition rate of around 44 packets per second with 512 data slots.
- ANSI E1.31, “Streaming ACN” (ESTA) and Art-Net (Artistic Licence). Transport of DMX data over IP networks.
- US patent 4,980,806, Taylor and Walsh, Vari-Lite Inc., granted on 25 December 1990: Computer controlled lighting system with distributed processing.
- Theatrecrafts, Backstage Heritage Collection: Vari*Lite Artisan and VL2. Secondary source, to be checked against manufacturer documents in the research dossier.
Research status. This page describes a research project in preparation. It does not describe a product or a standard. Hypotheses are labelled as such.
Important information on the free use of this text/paper: the free use of our text/paper requires that the reference to its origin and the corresponding link are not removed, altered, or replaced. This also applies to partial use of the content.
Wichtige Information zur kostenfreien Nutzung des Textes/Papers: Die Voraussetzung für die unentgeltliche Nutzung unseres Textes/Papers ist, dass der Hinweis auf die Herkunft sowie der entsprechende Link nicht entfernt, verändert oder ersetzt werden. Dies gilt auch bei der teilweisen Übernahme von Inhalten.
