MrNeRF (@janusch_patas)
2025-04-21 | โค๏ธ 97 | ๐ 13
Volume Encoding Gaussians: Transfer Function-Agnostic 3D Gaussians for Volume Rendering
Contributions: โข A high-quality, compressed representation (VEG) for structured and unstructured volume data using 3D Gaussians, enabling real-time rendering of large volumes with arbitrary transfer functions.
โข A direct initialization pipeline from volume data, eliminating the need for structure-from-motion scene reconstruction tools.
โข An opacity-guided multi-transfer-function training strategy, ensuring accurate reconstruction across an entire scalar field.
โข A comprehensive evaluation demonstrating that VEG achieves high reconstruction quality, extreme compression (up to 1800ร), and interactive visualization on large-scale scientific volumes.
๋ฏธ๋์ด
๐ Related
Auto-generated - needs manual review
Tags
domain-vision-3d domain-rendering domain-ai-ml domain-dev-tools domain-visionos