TerraNova Equestrian Center
PTZ cameras, IDFs, a couple of miles of fiber, two simultaneous live streams, and a gator. The broadcast system a former student and I have been building on a horse farm, and why the goal is for nobody to notice.
More than a decade ago, Dylan Billmaier was my student in a high school theatre program in Tampa. Today he's the Chief Technology Officer of TerraNova Equestrian Center, a 3,000-acre facility that hosts equestrian events all year, and for the last year and a half, I've worked for him.
I want to sit with that sentence for a second, because it's the best one on this page. Everything below it is cameras and fiber and network diagrams, and all of that is real and I'm proud of it. But the reason any of it exists is that a kid I taught grew up into someone who runs the technology for an entire property, looked around for help he could trust, and thought of me.
TerraNova's flagship events are cross country and combined driving competitions, and for years the video production and streaming for those was outsourced. Dylan wanted it in house. So he built it, and ran it himself for the first year or two. Then, in the spring of 2025, he called me. Not because he couldn't do it. Because he's the CTO of the whole property, show days are exactly when everything else on the property also needs him, and this was one hat too many.
So the brief, once we boiled it down, was a single sentence: on show day, Dylan walks into the production trailer, sees cameras up, and starts directing. He doesn't check a switch. He doesn't walk a fiber line. He doesn't go into the field during setup, because his normal job needs him elsewhere. If something breaks, he keys the radio, and it becomes my problem out in a field somewhere.
The inheritance
Dylan is not one to do things halfway. He knew this project was coming from the day he took the job, so by the time I arrived, the expensive, invasive, impossible-to-retrofit part was already done: fiber in the ground across the property, and a live MDF with the core switching already in it.
There was also a great deal of Kramer AV-over-IP gear, because TerraNova already uses Kramer encoders and decoders for distributed video everywhere on site: TVs in the pavilions, in the barns, wherever people need video signal. When Dylan needed to move camera feeds for a broadcast, the engineering decision was easy: the boxes already sitting on his shelf. I want to flag that as a decision, not a compromise. You start with what's in the building. Keep Kramer in mind, though. It comes back twice.
My first show onsite, I designed nothing. I shadowed Dylan and helped throughout the setup process. I ran the system exactly as it existed and kept a list: what worked, what didn't, what was going to fail eventually. Starting with the next show, the day-to-day became my responsibility: topology, switching, drawings, labeling, provisioning, and the physical deployment with a small team.
One thing to be clear about before the technical stuff starts, because the "I" in the paragraphs below can read as more solitary than the truth: this is a collaboration. The foundation, the decision to bring production in-house, and several of the biggest platform calls on this page are Dylan's. My job has been to take what he built, run it, refine it, and design what comes next with him, not for him.
One network, many tenants
There is no dedicated production network. Everything we deploy rides on the facility's actual working network infrastructure, the same one carrying guest Wi-Fi, corporate offices, and interoffice traffic, across the whole 3,000 acres.
The tool for the job is VLANs. If you've never met one: a VLAN carves a single physical network into several logically separate ones, so traffic that has no business meeting never does unless we deliberately route it. Production video gets a VLAN. Show Com gets a VLAN. Dante gets a VLAN. The offices and the guest Wi-Fi keep theirs. Same fiber, same switches, and nobody's Zoom call ever shares a lane with a camera feed.
The gear itself evolved across the shows: Cisco Meraki at the start to mostly Ubiquiti UniFi now. Meraki's licensing costs were eating the property's budget, Dylan was already moving the whole site to UniFi, and so production naturally followed. Students, take the lesson at face value: the best technical answer and the answer the organization can afford to keep running are different questions, and a designer who only ever answers the first one builds systems that die the year they leave.
The production Wi-Fi is its own SSID, broadcast only from the access points in the zones where I need it in the field.
One cable
When I arrived, a single camera position worked like this: a JVC PTZ with a network connection for pan-tilt-zoom control, plus an HDMI output into a Kramer encoder, which streamed over the network through the MDF to the trailer, where a Kramer decoder converted it back to HDMI to feed the switcher. Read that sentence again and count the things that can fail.
A camera position now works like this: a Panasonic PTZ with native NDI, and one network cable. Power, video, and control, all on the cable.
The obvious win is fewer points of failure, and setup time per camera fell from fifteen or twenty minutes to five or ten, which across nine cameras hands back about an hour and a half on every deployment day. But the number I actually care about isn't the ninety minutes. It's this: the upgrade changed who can deploy a camera.
Under the old workflow, handing a camera to a team member meant also handing them a signal flow diagram and trusting them to follow it exactly, because one wrong connection and we'd lose video or control with no obvious reason why. Under the new workflow, the entire briefing is "plug this cable into that port." Everything else happened on my bench days before the gear ever left the shop.
I wasn't optimizing the signal path. I was optimizing the workflow: for me, for the people helping me, and for whoever runs this after I'm gone. It's the same idea as the trailer brief, one layer down. The system shouldn't need Dylan in the field, and it shouldn't need me at every camera position either. A system that only works when the person who built it is standing next to it isn't finished.
Provisioning, in this scenario, meant: IP addresses assigned, switch ports configured with the correct trunk and access assignments, and the cameras themselves set with the right address and matched video settings so all nine cut together cleanly on a broadcast. Frame rate and resolution are set once and forgotten. Iris stays on auto, because this is Florida, the show runs all day, the light does what it wants, and most of our camera operators are students rather than career camera ops. The fewer parameters they own, the better they can do the one job that matters: follow the horse.
White balance stays mine. I match it across all nine and ride it through the day as the weather turns. Sometimes from a controller in the trailer during a break, since I mapped it to the user keys. More often from a web UI, either at the office nearly two miles away or from my laptop in the gator, because the site Wi-Fi reaches. There's a computer at the office with every web UI in the system already open in tabs: cameras, switchers, comms. When something needs changing, I'm not hunting an IP address, I'm clicking a tab.
Show day, inside the trailer
The trailer is where the show gets made. Dylan directs. Brett Lorins (another lighting human, actually!) runs the Roland switcher, and the PTZ operators sit in there with him, each on their own camera.
Every call is verbal, over comms. There's no rundown and no cue sheet, because… well… you can't cue a horse. After twenty years of building shows one cue at a time, there's something satisfying about a show that simply happens, live, and everyone in the trailer reacting to it.
The broadcast layer is OBS sitting on the switcher output, carrying our overlays and graphics, feeding Vimeo, which handles the cross-platform distribution from there. Two computers, one per stream, driven by Stream Decks.
Two streams, because during cross country there are frequently two shows at once: the course, and the dressage ring. All the cameras and all the complexity live on cross country. Dressage is one camera routed through a submix to the second machine. Dylan keeps half an eye on it, and when it goes to a break, Brett reaches over and presses one Stream Deck button that drops that stream to a placeholder or a commercial, then goes right back to working the real show.
And then there's the tally system, which is the smallest build on this page and the one I might be proudest of.
Brett always knows what's live, because the Roland's multiview has tally built in. The camera operators had nothing. No reliable way to know whether their camera was in preview or program, which means guessing or asking, and both of those cost you shots. So I built them one: GPIO straight off the Roland, driving the red and green channels of an RGB LED strip mounted under each operator's monitor. No microcontroller, no software. The Roland's GPIO closes a relay; the relay closes the circuit. Contacts touch, power flows, and every operator's monitor glows green when their camera is in preview and red when it's in program. Each operator gets a peripheral-vision answer to the only question they ever need answered, and there is nothing in the chain that can crash, update, or misbehave. It is the least sophisticated device in this system and it has never once let us down.
A half-mile away
The cross country control tent sits about a half-mile from the trailer. The commentators and judges work out of that tent, and the judges need live eyes on the entire course, because if a horse goes down or a jump breaks, they are the ones who stop the next horse from running. If the tent loses video, people can get hurt. That reality frames every engineering decision inside it.
The tent gets two 55-inch displays, one carrying the multiview with preview baked in, one carrying program, both fed by Kramer decoders tuned to their respective channels. (Told you Kramer would come back. We still encode multiview and program into the Kramer system precisely because it lets us put those pictures anywhere on the property that already has a decoder.) It gets Dante audio on its own VLAN, carrying the commentators' microphones out to the loudspeakers across the whole site, plus background music. Worth being precise here: that is site audio, for the people in the stands. The livestream has its own commentator in a separate booth. Two commentaries, two audiences, running simultaneously. The tent gets solid Wi-Fi, because the judges congregate there. It gets comms, and it gets redundant power: one generator running, one on standby, and a UPS in the rack so that if a generator dies, the room never dies while we swap over.
Prep, load-in, and the gator
The ratio is two to three days of prep to one day of deployment, and the ratio is the point.
Prep is paperwork, planning, and bench provisioning. Every device gets configured and labeled. Every position gets a Pelican case packed with exactly what that position needs: cables, adapters, fiber supplies, all of it. Every position also gets a cable-flow diagram, plus IDF-specific drawings, so the technician deploying it does not have to be inside of my brain. And every fiber patch in the MDF gets made before load-in starts, so that the moment we light up an IDF in the field, we know instantly whether the run is good instead of standing in the grass guessing. If you've read my Producers in the Park post, you already know this religion. Shop prep can make or break a show. Same doctrine here, just fiber and cameras instead of dimmers and Source Fours.
There were no site drawings, so we made our own: I pulled a Google Maps image, scaled it, and built an accurate working model from it. That model is where cable distances get calculated, IDF and camera positions get planned, and RF coverage zones get drawn. It is the identical drafting discipline I use on a light plot, aimed at a field. Between the buried backbone and the tactical runs, there's MILES of fiber in play on a show.
The IDFs are placed strategically, but strategically is never quite where a camera or a trailer wants to be, so tactical fiber runs along the ground from IDF to position, occasionally buried under a walking path, generally no more than nine hundred or a thousand feet. Each tactical run carries four strands, but we only need one (using BiDis). I patch two or three anyway, so that if a connector breaks or a strand comes up dirty mid-event, the fix is a hot-spare swap at the patch panel, not a field repair with horses running past.
Power is its own layer, because we haven't yet developed a practical way to transport electricity via light. Well, I guess lightning. But the switches wouldn't like that… Each IDF cabinet has a small generator powering the switch, which powers the camera. A generator is now a component in my signal chain, and knowing how to configure and maintain one is part of the job.
Some cameras sit on towers and platforms, which is easy. For the rest, I've fabricated custom brackets: sometimes hanging a camera off the underside of a boom lift, another off the bottom of a video screen. Those got designed and built on the spot with what's on hand, screws and wood, strut channel and bolts, whatever solves the geometry in front of me. Some are one-offs. Some worked well enough that they got refined and now come back every year. Simply put, this is stagecraft, pointed at a horse jump instead of a set piece.
Deployment day is me plus one or two people, with a bigger crew for loadout. I load the gator, which now has a cable rack I built onto its side carrying spares of everything, because I will forget something. Then I start at the farthest IDF and work my way home toward the MDF: camera up, position verified, next. In theory, by the time I reach the trailer, all pictures are up and waiting for me. In practice it's usually more than just one or two or four trips. An adapter got forgotten, a position needs something that wasn't in its kit, and a running list builds all day: this IDF needs X, that one needs Y.
And I'd be lying if I didn't tell you that the gator rides are half the reason I love this gig. There is something genuinely good about driving across an open field at six in the morning to turn cameras on, with the haze still sitting on the grass. I grew up in the country. I grew up outside. Most of my working life happens in a dark room in front of a console, and I love that room. But I'd forgotten how much I liked this, and it turns out I like it a lot.
Reading the symptoms
I teach troubleshooting harder than I teach almost anything else, so I want to be specific about how it works here instead of writing "and then I fixed it."
I can diagnose this system quickly for one boring reason: I helped design the signal chain, so I know exactly what it is.
From there, the symptoms sort themselves. If everything from a camera drops at once, video and PTZ control together, that is not a video problem; that's network or power, because all of it rides the one cable. If video is gone but the PTZ controller still talks to the camera, now it's a video problem and the network and power are fine. Then context clues: pull up the UniFi controller. Switch online but a port dark? The problem is whatever's plugged into that port. Switch offline entirely? Now it's the fiber path or the power, and only one of those is checkable from a laptop, because you cannot ping a generator.
And because the site Wi-Fi reaches, all of that diagnosis happens from the laptop on the drive out. By the time I'm standing at the position, two-thirds of the possibilities are already eliminated.
The honest footnote is that once we started burning the system in for a day or two before each event, mid-show network problems became rare.
Cameras in the jumps
The PTZs track the horses; that's the main part of the broadcast. But cross country has water jumps and genuinely inventive obstacles that the crew builds, and we wanted them on camera.
So we built four wireless GoPro rigs on Teradek transmitters and receivers. Each rig is a package: battery pack, transmitter, GoPro, rolled off the chargers each morning as a unit, with the camera unclamping from the pack to mount anywhere on SmallRig arms. Three of them are fixed, and because the rigs are tiny, we can hide them inside the jumps themselves, in spots that are safe for horse and rider and still get the shot. A horse coming over water from a foot away. It's so cool.
The fourth rig is a rover, and it's my favorite camera. One transmitter, three receivers pre-positioned around the course. Every hour or so I drive out and move it somewhere new, so the director always has a fresh angle sitting there for a toss shot or a picture-in-picture.
The RF got drawn before it got deployed: manufacturer specs turned into coverage circles on the scaled site model, giving rough distances between antenna positions. Then it got tested for real, me driving the rig around on the gator, watching signal strength on the Teradek and watching the picture itself for where it went choppy, finding the true handoff points. We have one enormous advantage, which is a giant open field with almost no RF interference, and the predictions held. A little refocusing, an obstruction cleared here and there, but usually the signal chain was solid.
We've also had a DJI drone on some events for aerial coverage, its receiver feeding, naturally, a Kramer encoder. No drone the last few shows, which is part of why the GoPros exist. If it returns, it replaces nothing; it's one more sweetener layer on top, and we'd bring it in over NDI this time.
Worth recording how the GoPro system began, because it didn't begin as a system. It began as one rig and a hodgepodge of parts, with me spending the gaps in a show day trying it in different spots to see what worked, knowing we wanted more rigs eventually.
That is most of what a show day is for me now. I find a shady spot near the center of the cross country loop, because once the course is live you have to look both ways for oncoming horses, which move FAST… and being central means I can reach anything fast without crossing much of the field of play, and I set up a little home base where I can watch the multiview for wireless dropouts and PTZ weirdness, and also just watch the horses go by. During breaks I'll pop into the trailer for the small stuff that isn't worth radio traffic, an operator who needs a sensitivity tweak on their controller but is too locked in to deal with it, or honestly just to sit in the air conditioning with my friends for a few minutes. And the whole time, some part of my brain is designing the next show inside this one. That camera's fine, but it wants to be twenty feet down the road. That GoPro shot worked, but it'd be better a little lower to the water.
Around the property
The point
TerraNova is not theatre, yet I want it on this website anyway. I do lots of little side things like this, and it's my hope in the coming months/years to bring more of these narratives over to the site.
Most of what people know me for is lighting design, and I'd like the students I teach, and anyone else wandering through this site, to see that the career is not only making pretty light. It's networks and RF and fabrication and generators and standing in a wet field at six in the morning solving a problem no textbook contains. All of it ties loosely back to the production world I live in, and all of it is also very much its own thing, and moving between those worlds is a large part of why my career looks the way it does.
But if one thing on this page sticks, I'd rather it not be the technical part. The technical knowledge is the cheapest thing here. It arrives with time, repetition, curiosity, and above all with being the kind of person other people want in the room. The expensive things are the rest: the prep, the process, the discipline of designing for whoever comes after you, the habit of watching this show while quietly building the next one, and the relationships that put you in the field in the first place. I'm at TerraNova because a student I taught more than a decade ago grew into someone who trusted me with his site… No certification or training program covers that.
Which brings us back to the original mission. On the best show days, the deployment held, the burn-in did its work, and the radio on my hip stays quiet. Dylan never leaves the trailer. Operators never ask where their tally went. The judges' tent never goes dark. And I sit in the shade at the center of the course with a laptop I don't need to open, watching the horses run.