Calibration without a physical bar & automatic gaussian-splat loops

(preview)

Calibration Without A Physical Bar

The new 2026 calibration workflow in XangleCS removes one of the biggest setup headaches: you no longer need a dedicated physical calibration bar to get started.

Instead of stopping the session to capture a separate bar reference, you can use a regular reference shot from the rig and let XangleCS calculate the calibration for you. The initial solve takes about a minute, but once it is done, that calibration can be reused across the rest of the shots from the same setup, just like the older fast workflow. The difference is that the setup is lighter, the process is more flexible, and the operator has one less physical accessory to manage on location.

That changes the feel of the whole capture process. It is easier to move quickly, easier to work in tighter conditions, and easier to keep momentum when you are trying to get a usable dataset without breaking the rhythm of the shoot. For teams running multi-camera captures under real production pressure, removing a dedicated calibration prop is not a small detail. It reduces friction at exactly the stage where delays and uncertainty usually start to stack up.

We also added new visualization tools to make the calibration easier to understand and easier to trust. XangleCS now helps you identify problematic cameras earlier, see where a rig may be drifting or misbehaving, and review a clearer preview right away instead of waiting until much later in the pipeline to discover that something is off. That means faster troubleshooting, better decisions during setup, and fewer unpleasant surprises once you move into processing.

The result is a calibration workflow that feels more practical in the field: less hardware, less interruption, faster diagnosis, and a much clearer path from capture to a stable result.

From Calibration To Gaussian Splat Playback

Once the rig is calibrated, XangleCS can take the capture much further. You can generate a Gaussian splat .ply file from the dataset and then render moving camera paths from that result, either automatically with presets or manually with more direct control.

That second part matters more than it may seem. One of the hardest problems in automated Gaussian splat playback is not just building the splat itself. It is choosing viewpoints that actually feel intentional. A technically correct reconstruction is only part of the job. If the camera path feels awkward, random, or disconnected from the original rig, the finished playback still will not read well.

XangleCS is designed to help solve that. Presets give you a fast way to produce repeatable moves, which is useful when you want speed, consistency, or quick proofs. But when you want a more deliberate result, the sequencer lets you build the playback from specific camera points of view and shape the motion more intentionally. That gives you a much better way to control how the subject is revealed, where the movement begins and ends, and which viewpoints carry the strongest visual impact.

In practice, that means you can move from a calibrated dataset to a renderable result without treating playback as an afterthought. Start with automation when you want momentum. Switch to the sequencer when you want to refine the camera path into something that feels more authored. The goal is not only to output a Gaussian splat file. It is to turn that file into a playback that feels stable, readable, and worth showing.

From quick tests to more polished hero renders, Orbit gives you a practical workflow for turning multi-camera capture into directed Gaussian splat playback.

The new functionalities are available now within a side app called Xangle Orbit (studio version to re-process your older shots and make cool demos). Hit me up if you’re up to try this one! The complete functionalities should be available in a few weeks in the new XangleCS