用扩散模型生成符合物理规律的非负三维云体，速度快且逼真。
NnD: Diffusion-based Generation of Physically-Nonnegative Objects
- 基于得分函数的扩散模型，通过设计保证生成过程全程非负。
- 生成的3D云体与真实云物理趋势一致，专家难辨真伪。
- 适合需要快速生成高保真物理场景的研究者或开发者。
多数自然物体具有内在复杂性和变异性。尽管一些简单物体可从第一性原理建模，但诸如云形成等真实现象需计算量巨大的模拟，限制了可扩展性。本文聚焦一类在物理上合理、非负且计算上可行但模拟成本高的对象。为大幅降低计算开销，提出非负扩散（NnD）：一种基于得分函数的生成模型，采用退火Langevin动力学，在迭代生成和分析过程中显式强制非负性。NnD在高质量物理模拟数据上训练，训练完成后可用于生成与推理。我们展示了3D体积云的生成，其包含固有的非负微物理场。生成结果与云物理趋势一致，专家感知测试中未被识别为非物理。
原文摘要 · Abstract (English)
Most natural objects have inherent complexity and variability. While some simple objects can be modeled from first principles, many real-world phenomena, such as cloud formation, require computationally expensive simulations that limit scalability. This work focuses on a class of physically meaningful, nonnegative objects that are computationally tractable but costly to simulate. To dramatically reduce computational costs, we propose nonnegative diffusion (NnD). This is a learned generative model using score based diffusion. It adapts annealed Langevin dynamics to enforce, by design, non-negativity throughout iterative scene generation and analysis (inference). NnD trains on high-quality physically simulated objects. Once trained, it can be used for generation and inference. We demonstrate generation of 3D volumetric clouds, comprising inherently nonnegative microphysical fields. Our generated clouds are consistent with cloud physics trends. They are effectively not distinguished as non-physical by expert perception.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。