Volumetric reconstructionTurn camera views
Turn camera views
into a 3D performance.
We are developing software to rebuild a moving person from cameras that recorded them at the same time.
Research & development
A multicamera array records a performance from different viewpoints. Our solver development examines how to reconstruct the moving subject while preserving its shape and consistency over time.
We work on the relationship between camera calibration, depth and the resulting spatial representation. The aim is reusable reconstruction software for capture teams and spatial media developers.
- Designed for
- Volumetric capture teams and spatial media developers
- Input
- Synchronized camera views, calibration and depth
- Target output
- Moving three-dimensional subjects for spatial playback
- Current focus
- Depth alignment, camera geometry and temporal consistency.
8 images
21 imagesEngineering context
Camera geometry, depth and playback.
Our research connects calibrated views, estimated depth and synthetic scene data with the renderer that presents the result.
Explore the workTalk to us about the technology.
Discuss capture systems, GPU imaging or spatial playback development.