Gracia's groundbreaking volumetric capture technology is revolutionizing the immersive experience, and its recent advancements in streaming are particularly exciting. The company's journey began with a Quest app for static photorealistic objects, but its true potential lay in moving volumetric scenes, a feat achieved through Gaussian splatting. This technique involves fitting millions of semitransparent colored blobs in 3D space, allowing for realistic rendering from any viewpoint in real-time.
The initial standalone headset app faced a significant challenge: the need to download multi-gigabyte files for each scene, requiring a 2.4 gigabit internet connection, which is not feasible for most users. This limitation hindered Gracia's vision of becoming the 'YouTube of truly volumetric content'. However, the startup persevered, and after a year and a half, it achieved a remarkable breakthrough.
In March of this year, Gracia launched a world-first: the ability to stream select 4DGS scenes without any app install or full scene download. This innovation is made possible by encoding only the keyframes and motion change deltas, rather than sending the entire scene for each frame, a technique akin to video codec approaches. The result is near-instant streaming of volumetric captures, with no hard file-size cap, opening up possibilities like entire concerts.
The streaming quality requires a constant 75 Mbps connection, which is now accessible to around 90% of homes in developed countries. However, for those with slower connections, Gracia offers a 17 Mbps mode, albeit with a significant quality hit. Three scenes are currently available to stream, showcasing a person working on a bicycle wheel, a doctor examining a patient's shoulder, and an impressive 4-minute performance by a musician singing with a guitar, complete with spatial audio.
Gracia's technology is not limited to mixed reality; it can also be placed into existing virtual Unity or Unreal environments, including for VR games. The capture and processing requirements are substantial, involving around 60 shutter-synced and genlocked cameras, but the results are truly remarkable. I, for one, would much rather watch a musician or comedian perform as a 6DoF volume in my living room than a 3DoF 180° 3D video.
The next frontier for the volumetric scenes space is developing technology to allow anyone to capture volumetric scenes with just a couple of iPhones and significantly reduce processing time. While this may seem impossible today, the pace of progress in 3D reconstruction, driven by AI advancements, is staggering. The 'YouTube of truly volumetric content' may not be possible today, but it will likely become a core use case of XR technology by the end of this decade, propelling its mainstream adoption.
