Research draft · 3 November 2025

ZentrumNeueLichtkultur — Research · SRLC

How it could work

JoJo Tillmann · ZNL ·

Distribute once. Trigger together. Execute locally.

This page shows a draft. No such control system exists, neither as a product nor as a standard. The page explains in plain words how it is meant to work and lets you try out in a simulation what changes compared with today’s data stream, and what does not.

The idea in three sentences

The fixture knows its part

Today the console calculates every transition and sends the result to the rig, value by value. In the draft, each device receives its part of the show in advance and carries it out itself on a common signal.

1 · Distribute

The data moves house

The console sends each device its cues: target values, times, curves. Once, secured, with a version number.

2 · Trigger

One signal for all

For the cue, a short message goes out to all: cue 12, start at this moment. All devices read the same clock.

3 · Report back

The console knows what is running

Every device confirms and reports its state. If a report is missing, the console shows it.

Much like an orchestra: the parts are on the music stands, the conductor gives the cue. He does not call out every note to every musician. For that, everyone needs the same edition of the score, and anyone who wants to improvise needs another way.

Three models

Where execution lives

The three models differ in who holds the data and who calculates the transition. None of them is presumed the winner.

Model A

In the fixture

Fixture Intelligence. Every device stores and calculates for itself.

Strength
Keeps running if the network fails. The transition is calculated where the motor sits.
Limit
Every device needs a network connection, a clock and memory. A replacement unit has to load first.
Selectable in the simulation
Model B

In the node

Distributed Edge Control. A node runs a group of conventional fixtures.

Strength
Works with existing DMX devices. Replacing a fixture stays simple.
Limit
Behind the node everything stays as it was, including the missing return channel. If the node fails, the group fails.
Selectable in the simulation
Model C

In the meaning

Semantic Lighting Network. The network transmits tasks instead of values.

Strength
“Soloist position, cool white, in 3 seconds” applies to every type of device. Changing the type needs no reprogramming.
Limit
Two manufacturers have to mean the same thing by “cool white”. The result is harder to predict.
Example message only, not simulated

The data stream stays.Anyone pulling a fader or focusing a fixture by hand needs values in real time. A hybrid model is therefore likely: stored sequences for the show, a live channel for the hand. Where the line runs is a research question.

Try it out

One cue, three ways

Choose an architecture and a rig size, trigger a cue and cut the network in the middle of the transition. The transition takes 4 seconds.

SRLC simulation · draftruns in your browser · example values · not an existing product
Architecture
Fixtures
Diagram: lighting console, switch, four nodes and fixtures with animated data flow
What goes over the network
    Console → rig since the last cue
    0Messages
    0Data volume
    Where the data lives

    The dots stand for fixtures; in large rigs one dot stands for several. Pink dots on the lines are triggers, white ones are continuous values, green ones are feedback. The simulation works with fixed example assumptions, listed below the comparison table. It measures nothing.

    Technical comparison

    What changes, in numbers

    The table calculates for the rig size chosen above. Note the second row: bandwidth is not the argument.

    What the numbers do not show: the data stream is wasteful but robust. If a packet is lost, the next one arrives 23 milliseconds later. A single trigger must not be lost. It needs repetition, confirmation and a rule for the case where both fail to arrive.

    Example messages

    What is on the wire

    Four messages are enough to show the principle. The notation is chosen for readability. A real system would use a compact binary format such as CBOR.

    Messages · draftexample schema · not a standard
    1 · Distribute: cue data for one fixture
    {
      "type": "cue_data",
      "show": "Sample_Tour",
      "version": 7,
      "device": "spot-042",
      "cue": 12,
      "target": { "dimmer": 0.80, "pan": 112.5, "tilt": -34.0, "colour_xy": [0.31, 0.33] },
      "dauer_ms": 4000,
      "curve": "sine",
      "checksum": "sha256:9f2c…"
    }

    Sent to every device in advance, secured. The version number prevents devices running different versions from playing the same show.

    2 · Trigger: one signal for all
    {
      "type": "go",
      "show": "Sample_Tour",
      "version": 7,
      "cue": 12,
      "start": "T + 50 ms"
    }

    Sent as one message to all, several times to be safe. “T” is the shared clock. The start time lies slightly in the future so that network latency does not matter. This lead time is the price: the cue comes a few milliseconds after the key press.

    3 · Report back: the state of one device
    {
      "type": "status",
      "device": "spot-042",
      "version": 7,
      "cue": 12,
      "phase": "running",
      "clock_offset_us": 38,
      "errors": []
    }

    The console sees who is running which cue with which version, and who is silent.

    4 · Model C: a task instead of values
    {
      "type": "task",
      "group": "front-spots",
      "target": { "position": "soloist", "colour_temperature_k": 5600, "intensity": 0.8 },
      "dauer_ms": 3000,
      "start": "T + 50 ms"
    }

    The device decides for itself which motors and LEDs to drive, and how. This assumes that it knows its position in the room and its own capabilities.

    What it takes

    Eight building blocks, most already there

    The draft does not invent a transport protocol. It asks which standardised building blocks are enough and where a gap remains.

    No.Building block
    In the device or node
    1Storage for sequence dataThe cues, times and curves of a show, with version and checksum.
    2Local calculationThe device calculates transitions itself, within its own movement and operating limits.
    3Defined behaviour without a networkWhat happens when the console falls silent: hold, finish, or go to a safe state. The show decides this, not chance.
    In the network
    4Shared clockTime synchronisation over Ethernet is standardised (PTP, IEEE 1588). The accuracy a lighting rig needs has yet to be measured.
    5Reliable deliveryTriggers to all, confirmations back, repetition on loss.
    6Message authenticityA device that acts on its own must know who a command comes from.
    Across manufacturers
    7Common data modelA description of cue, target state and curve that every device understands in the same way. Device descriptions such as GDTF are a starting point.
    8Independent stopA way to stop all movement that does not depend on the control system. The more autonomous the device, the more this matters.

    What already exists

    • Ethernet right into the fixture, on many current devices.
    • Standardised time synchronisation, in everyday use in audio and video.
    • Return channel and device management via RDM and RDMnet.
    • Open device descriptions (GDTF) and scene exchange (MVR).
    • A historical precedent: the Vari-Lite Series 200 system from 1987.

    What is still open

    • How do you play live when the sequences live in the device?
    • How quickly does a replacement unit learn its part in the middle of a show?
    • How does a rig get back in step once the network returns?
    • Who is liable when a device completes a movement without a console?
    • Why would manufacturers want a common data model?

    Cybersecurity, operational safety and functional safety are three separate questions. The draft answers none of them conclusively.

    Your part

    Tell us where it sticks

    A draft gets better when practitioners take it apart. We are most interested in what the simulation is missing or weights wrongly. And we are looking for people who have worked with Artisan and Series 200.

    Objections, experience, measurements? Write to us:

    srlc@zentrumneuelichtkultur.de

    To be honest

    What the approach could do, and what it cannot

    It could

    • Finish a running transition, even if the network fails.
    • Show the console what the rig is actually doing.
    • Calculate transitions where the motor and LED sit, at the device’s own rate.
    • Free shows from the channel layout of individual device types.

    It cannot

    • Replace the data stream. Live playback and focusing need values in real time.
    • Solve a bandwidth problem. There is none.
    • Be safer by itself. A device that keeps moving without a console is a risk too.
    • Work without agreement between manufacturers. That, we assume, is where the 1987 precedent failed.

    Status

    Where we stand

    DoneResearch question, project brief, project page, this simulation as a first draft.
    In progressResearch dossier with sources: history, protocols, distributed real-time systems. Then the white paper.
    OpenMeasurements on real hardware, a prototype, talks with manufacturers and standards bodies. We will only name dates once they are fixed.

    The simulation figures follow from the stated example assumptions and from the packet structure under ANSI E1.11 (DMX512-A) and ANSI E1.31 (sACN). The information on Vari-Lite is based on US patent 4,980,806 and is being checked further in the research dossier.

    The ZentrumNeueLichtkultur is a non-profit organisation, independent and free-thinking. It started in 2020 as a regional phenomenon and is now an international network for lighting professionals. SRLC is a research project in preparation.

    Important information on the free use of this text: the free use of our text 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: Die Voraussetzung für die unentgeltliche Nutzung unseres Textes 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.