Zhuma Innovation Funds 3DGS Spatial Camera

🔑 Enhanced Key Takeaways

• •Zhuma Innovation's Pebble utilizes a proprietary multi-sensor fusion algorithm that specifically optimizes 3D Gaussian Splatting (3DGS) for mobile processing, reducing the reliance on high-end desktop GPUs for initial scene capture.
• •The company is positioning its 'space memory camera' as a direct competitor to existing spatial capture solutions by integrating a cloud-based rendering pipeline that allows users to share interactive 3D scenes via standard web links without requiring specialized viewer software.
• •The funding round is specifically earmarked for the development of an open-source SDK that will allow third-party developers to integrate Zhuma's 3DGS reconstruction engine into existing AR/VR mobile applications.

🛠️ Technical Deep Dive

• •Architecture: Employs a hybrid approach combining sparse LiDAR point clouds for geometric constraints and 3D Gaussian Splatting for high-fidelity texture and lighting representation.
• •Processing: Implements a custom CUDA-accelerated kernel for mobile chipsets to enable real-time splatting preview at 30fps during the scanning process.
• •Data Format: Outputs proprietary compressed .gs files that support progressive loading, allowing for streaming of high-resolution 3D assets over standard 5G/Wi-Fi networks.
• •Sensor Suite: Utilizes a synchronized dual-camera setup for stereo depth estimation, supplemented by an IMU for high-frequency motion tracking to minimize drift during handheld capture.

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline