发表机构
Princeton University（普林斯顿大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
SCION提出层次化组合场景表示，用可复用基元和轻量级实例替代独立高斯，实现紧凑、可控、可编辑的神经场景表示。
AI 中文摘要
真实世界场景具有组合性：砖块、草叶、鹅卵石和树叶在人工和自然环境中反复出现。现有的神经场景表示将这些元素独立建模。大多数3D高斯溅射及其后续的抽象和压缩方法将每个元素视为唯一，为每个场景拟合数百万个独立高斯。先前的方法如Splat和Replace拟合模板对象，但需要手动选择重复元素。因此，这些表示存储冗余参数，并为下游任务提供弱操作句柄。我们引入SCION，一种层次化组合场景表示，用紧凑的可复用基元词汇表和轻量级世界空间实例替换独立高斯，这些实例在整个场景中放置变换后的副本。我们通过离散和连续场景参数的联合优化来拟合多视角捕获，结合了两级致密化（针对溅射和实例）和对抗性损失，以保留共享基元间的细节。恢复的结构产生紧凑、可控的表示，即使在1.2 MB下也保持高质量。SCION实现了有利于现有高斯压缩方法的率失真，并支持无需重新训练的实例级场景编辑和动画。我们的结果表明，神经场景表示无需将场景记忆为独立基元；它们可以发现可复用部分。项目网页：此https URL
英文摘要
Real-world scenes are compositional: bricks, blades of grass, pebbles, and tree leaves recur across human-built and natural environments. Existing neural scene representations model these elements independently. Most 3D Gaussian Splatting and follow-up abstraction and compression methods treat each element as unique, fitting millions of independent Gaussians per scene. Prior methods like Splat and Replace fit template objects, but they require mostly manual selection of repeated elements. As a result, these representations store redundant parameters and provide weak manipulation handles for downstream tasks. We introduce SCION, a hierarchical compositional scene representation that replaces independent Gaussians with a compact vocabulary of reusable primitives and lightweight world-space instances that place transformed copies throughout the scene. We fit this representation to multi-view captures via a joint optimization over discrete and continuous scene parameters, combining two-level densification over splats and instances with an adversarial loss that preserves detail across shared primitives. The recovered structure yields a compact, controllable representation while maintaining high quality even at 1.2 MB. SCION achieves rate-distortion favorable to existing Gaussian compression methods, and it enables instance-level scene editing and animation without retraining. Our results show that neural scene representations need not memorize scenes as independent primitives; they can discover reusable parts. Project webpage: https://light.princeton.edu/SCION
CommentsAccepted to NeurIPS 2026
