Camera tracking is at the heart of any real-time compositing pipeline. Without accurate, low-latency positional data, the virtual world and the physical world fall apart — you get the dreaded swimming effect where the virtual background drifts away from where it should be relative to the camera. At Vedri, we’ve built our tracking pipeline around the Mo-Sys StarTracker Max, and in this post we’re going to walk you through how it works, how we’ve integrated it into our setup at Vedri North Wales, and why we think it’s the right choice for a studio of our scale.
How the StarTracker Works
Despite the name, Mo-Sys StarTracker doesn’t actually track stars. It tracks a constellation of small retroreflective dots — roughly 3–4mm adhesive stickers — applied to the studio ceiling in a randomised pattern. A small sensor unit mounted to the camera lens looks up, captures the position of those dots, and uses the unique spatial arrangement of the pattern to triangulate the camera’s exact position and orientation in 3D space.
The result is six degrees of freedom — X, Y, Z position plus pan, tilt, and roll — delivered at low latency over a standard network connection. That data is then fed directly into Unreal Engine via the Mo-Sys M·Path plugin, which drives the virtual camera inside the engine to match the physical camera’s position precisely, frame by frame.
What sets the StarTracker Max apart from older optical tracking systems is its extended tracking volume and improved robustness under different lighting conditions. High-intensity studio lighting can confuse some optical systems — the Max handles it well, maintaining lock even with aggressive key and fill lighting setups.
Our Setup at Vedri North Wales
We installed the reflective dot grid across the full ceiling of the main studio volume at Vedri North Wales. The calibration process took a single afternoon — a Mo-Sys engineer walked us through surveying the space, generating the dot map, and feeding that into the tracker firmware. Once calibrated, the system just works. We’ve had the same calibration running reliably for months without drift.
The StarTracker unit itself mounts to a standard 15mm rod rail on the camera, feeding data through a small receiver box that sits on the rig. On our main camera — a Blackmagic Pyxis 6K — the integration is clean. The lens encoder attaches to the follow focus motor and feeds focal length and focus distance back into the system alongside the positional data, giving the engine full lens metadata to drive depth of field and perspective matching accurately.
Data flows from the StarTracker receiver into a dedicated tracking PC running Mo-Sys M·Path, which talks to our Unreal Engine render node over a 10Gb local network link. Latency from physical camera movement to background render update is consistently under 2 frames — imperceptible on set, and well within the range where talent and directors stop noticing the technology and start engaging with the environment.
Why Camera Tracking Matters More Than People Expect
If you’re new to virtual production, it’s tempting to think that the display technology is the hard part. It isn’t. Getting the tracking right is the hard part. A pixel-accurate virtual environment will look broken the moment the camera moves, if the tracking data is even slightly wrong. Parallax shifts, scale mismatches, and latency all read immediately on camera — and audiences are surprisingly sensitive to it, even if they can’t articulate why.
Good tracking means your director of photography can move the camera exactly the way they would on any other shoot — handheld, on a dolly, on a Steadicam — and the virtual world responds convincingly. It frees the creative team from the constraints of locked-off or pre-plotted moves that older compositing techniques required.
What’s Next
We’re currently testing integration with a handheld rig — running the StarTracker on a shoulder mount with a lightweight zoom lens. The challenge is vibration isolation; high-frequency camera shake can introduce jitter into the tracking data if the sensor isn’t properly damped. We’ll write up our findings once we’ve settled on a rig configuration that works cleanly.
If you’re exploring camera tracking options for your own virtual production setup, feel free to book a studio visit — we’re happy to walk you through the system in person and talk through what works at different scales.
