360° video is immersive by design—but delivering it at high quality is harder than it looks. When spherical footage is flattened into a standard video file, every part of the scene gets treated equally. That's a problem, because viewers don't watch it that way.

Dolby researchers tackled this head-on with a mesh-based spatial reshaping framework—a perception-aware method that allocates more pixel detail to the regions viewers are most likely to focus on and less to areas they rarely notice.

Smarter Pixels: Making 360° Video Look Better Where It Counts

The core idea is elegant: a lightweight mesh maps the spherical video onto a 2D plane, with vertices adjusted based on visual saliency. Important regions expand; peripheral ones compress. The result is a smarter use of every available pixel—without changing how the video was captured.

Crucially, the system is fully reversible and GPU-friendly. Backward reshaping during playback is handled using standard mesh rendering pipelines already supported by modern hardware, enabling real-time viewing with no custom infrastructure required.

In testing, the approach delivered sharper, more faithful reconstructions in salient regions compared to traditional projection methods—while fitting cleanly into existing 360° video compression and delivery pipelines.

As immersive media grows across education, tourism, and interactive experiences, this research points toward a future where visual quality is guided not just by geometry but by how people see it.

Dolby's Perception-Aware Approach to 360° Video

{{ pageEducationalSlot.headingContent[index] }}