Technology
How Do You Control LED Strips Over DMX and Art-Net?
DMX512 was designed decades ago to control stage fixtures, not thousands of individually addressable pixels. That mismatch is the root of almost every problem integrators hit when they set up DMX LED strip control on a modern pixel installation.
This article explains where the limits come from, how Art-Net extends DMX over Ethernet, and when it’s smarter to drive pixels directly over SPI. It’s written for integrators, event technicians, and installers who plan pixel projects rather than single fixtures.
What Is DMX512 and Why Does It Cap Pixel Counts?
DMX512 is a serial control protocol in which one universe carries exactly 512 channels. It was invented to control lights on increasingly complex music stages, and it still does that job well. The design assumption was a moderate number of fixtures, each using a handful of channels — not a strip where every LED is its own fixture.
That assumption is what breaks down on pixel installations. A pixel isn’t a light you address once, but one you address continuously alongside hundreds of its neighbors. The protocol never ran out of usefulness, only out of channels.
Why Only 170 Pixels per Universe?
One RGB pixel needs three channels — red, green, and blue. Since 512 doesn’t divide by three, the closest usable figure is 510 channels, which gives you 170 controllable pixels per universe. Two channels per universe are left stranded at the end.
To go past 170 pixels you need a second universe, which adds another 512 channels. On a 5m roll with 300 pixels, a single strip already spans two universes.
Which Channels Should You Leave Unused?
It’s considered bad practice to use the last two channels of each universe. The 171-pixel case shows why. Your first 170 pixels fill channels 1–510, channels 511 and 512 sit idle, and the remaining pixel has to start in the next universe.
You have two options. Either begin that pixel at channel 1 of universe 2, or split it across channels 511 and 512 of universe 1 plus channel 1 of universe 2. The second works, but it makes the patch hard to maintain.
How Does Art-Net Extend DMX Over Ethernet?
Art-Net is a protocol that carries multiple DMX universes over a single Ethernet connection. It was designed to give access to many universes from one device, and stage designers adopted it quickly. Today, every professional control system lists Art-Net among its control options.
The core win is cabling and scale. Instead of a separate cable per universe, you combine universes and send the data over one Ethernet run. Official numbers put the maximum at well over 32,000 universes on a single network — a ridiculously high number of RGB pixels in practice. An Art-Net decoder like SPI Matrix takes that Ethernet feed and turns it back into pixel data on its outputs.
Like DMX before it, Art-Net is a standard rather than a vendor format. Different manufacturers and software developers build against the same specification, which is why a pixel controller from one brand talks to a console from another. Nothing is free, though — depending on your software or hardware, you typically pay for every universe you enable.
Where Does Art-Net Inherit DMX Limits?
Art-Net is, in the end, DMX over Ethernet. Every DMX512 constraint travels with it: 512 channels per universe, 170 RGB pixels per universe, and the same convention of leaving the last two channels alone.
The real friction appears when you assign universes to individual outputs. Without an unlimited budget you’ll end up splitting one universe across several outputs, and the addressing stops being obvious.
Project Example: How Do You Map 600 Pixels to Universes?
Take two rolls of 5m, 300 pixels each, on two separate outputs. The pixel count is trivial; the addressing is not. This is the calculation that catches people out on site.
Output 1 uses universe 1 for its first 170 pixels and universe 2 for the remaining 130. That leaves 120 usable channels in universe 2, or 40 pixels. Wasting them isn’t an option on a budgeted project, so output 2 starts in universe 2 at channel 391.
| Output | Universe | Start channel | Pixels | Remaining in universe |
|---|---|---|---|---|
| 1 | 1 | 1 | 170 | 0 usable |
| 1 | 2 | 1 | 130 | 120 ch / 40 px |
| 2 | 2 | 391 | 40 | 0 usable |
| 2 | 3 | 1 | 170 | 0 usable |
| 2 | 4 | 1 | 90 | 240 ch / 80 px |
Four universes cover 600 pixels, and universe 4 still holds 80 spare pixels for a future run. Every one of those start addresses has to be entered correctly in the console, the software, and the controller. Change one roll length later, and the whole patch shifts.
Art-Net vs SPI: Which Output Should You Choose?
The alternative is simpler than it sounds: use the SPI output directly, with no Art-Net involved. You take the 300 pixels and assign them to an SPI output. No addresses, no universes — you think in pixels instead of channels.
| Criterion | Art-Net (DMX over Ethernet) | Direct SPI output |
|---|---|---|
| Addressing unit | Universes, channels, start addresses | Pixels |
| Pixels per universe | 170 RGB | Not applicable |
| Unused channels | Last two per universe | None |
| Licensing | Often paid per universe | Included with the controller |
| Industry compatibility | Standard, works with most consoles | Controller-side |
| Network load | Higher, sent continuously | Roughly 3× lower in our test |
| Patch complexity, multi-output | High | Low |
We measured the network side on the roughly 2,000 pixels used in our studio background. The SPI transfer sat around three times lower than the equivalent Art-Net data. The gap comes from how Art-Net behaves, not from the pixel count.
Art-Net data is sent out to the network whether or not the device is present. That’s by design — broadcast lets you feed the same data to several listening devices at once. On very large projects the same feature can clog the network, so watch for bottlenecks at routers and Ethernet switches.
Can You Combine Art-Net Triggers With SPI Output?
Yes, and this is the setup we use most often. You keep SPI for the pixel data and use Art-Net purely as a trigger layer, which preserves compatibility with most control systems in the industry.
Predefined channels trigger animations from an SD card or from running software, and control aspects such as speed and brightness. The show desk still speaks Art-Net; the pixels are still addressed as pixels in LED Strip Studio 4.
Source: LED Strip Studio
Everything above assumes addressable pixels. If your project uses plain analog RGB strips instead, DMX still applies, but the hardware is different—see our guide on how a DMX decoder for LED strips works.
Which DMX and Art-Net Mistakes Cost You Most?
Most DMX LED strip control problems are addressing and network mistakes, not hardware faults. These are the ones that repeat.
- Using the last two channels of a universe. It breaks the convention and makes the patch unreadable to the next technician.
- Sizing a project in pixels but patching in channels. Decide the addressing scheme before you cut a single roll.
- Choosing Art-Net when nothing on site needs it. If no console is involved, SPI removes the entire universe layer.
- Forgetting that universes are licensed. That cost is often discovered after the hardware is already bought.
- Ignoring switches and routers. Broadcast traffic saturates cheap network gear long before the controller becomes the limit.
- Skipping the pixel map. Without pixel mapping software, effects can only run along the cable, not across the room.
Planning a pixel installation and unsure whether you need universes at all? Book a free consultation with our team.
Summary
One DMX universe carries 512 channels, which equals 170 RGB pixels, and the last two channels should stay unused. Art-Net is DMX over Ethernet — it removes the cabling limit, not the channel limit, and is usually licensed per universe. Splitting universes across multiple outputs creates start-address arithmetic that is easy to get wrong on site. Direct SPI output lets you address pixels as pixels, and measured roughly three times lower network traffic on a 2,000-pixel install. Art-Net still earns its place as a trigger layer for animations, speed, and brightness, even when SPI carries the pixel data.