We have been working with Main Course Films on producing technical workflows for the AVP.
As Apple Immersive Video (AIV) production starts to move beyond experiments and into real-world delivery, one question keeps surfacing: how do you actually get high-end immersive content onto streaming platforms without drowning in data?
Working with Main Course Films, this video shows a new approach and testing of a practical delivery pipeline using the short film The Dobos Connection as a case study. The goal here wasn’t theoretical; it was to find a workflow that could hold up under real production constraints while maintaining the image fidelity that immersive storytelling demands.
This new pipeline is designed specifically for streaming delivery, not for every stage of the ecosystem. You still need AIVU for local playback and previews on the Apple Vision Pro, and ProRes mezzanine bundles remain essential for long-term archive. But for getting content to streaming services, this approach hits a compelling middle ground.
Instead of relying on heavyweight ProRes packages or lower-quality preview encodes, the workflow leverages tools already available in DaVinci Resolve to generate a high-quality intermediate using MV-HEVC. The result is a source file that is dramatically more efficient, roughly one-tenth the size of a ProRes bundle, while still retaining the fidelity needed for downstream HLS encoding.
On Director Clara Chong’s film The Dobos Connection, that translated into a very practical difference: an 83 GB upload instead of 1.3 TB. In other words, you’re reducing the upload to roughly 1/16th the size, which is exactly why your workflow feels so dramatic in practice. At scale, that’s not just convenience, it’s the difference between feasible and prohibitive.
At the core of the workflow is MV-HEVC (H.265), pushed well beyond consumer playback constraints. The material is delivered at full 8160 x 7200 resolution, 90 fps, stereoscopic, with data rates reaching up to 2,000 Mbps. That’s far above the ~150 Mbps typically required for playback in visionOS, but that’s exactly the point, this approach is all about preserving quality before the final adaptive streaming encode.
Critically, delivering at this full resolution enables the pipeline to leverage SpatialGen’s downscaling, which is tuned specifically for Apple’s immersive format. In practice, that produces noticeably cleaner results than encoding directly at playback resolution, particularly in fine detail and motion stability.
The use of MV-HEVC also sits in a useful middle ground: far more efficient than ProRes, but capable of sustaining much higher resolutions and data rates than anything intended for direct playback on current headsets.
There’s also a forward-looking element here. There have long been rumours of a higher-quality immersive encoder coming to Resolve. If and when that lands, workflows like this should only improve. But even today, this approach represents a meaningful step up in both image quality and delivery efficiency over existing options.
For now, it’s a pragmatic solution: one that acknowledges the realities of current tooling while pushing as close as possible to the limits of immersive image fidelity.