English

Geometric Decoupling: Diagnosing the Structural Instability of Latent

Computer Vision and Pattern Recognition 2026-04-22 v1 Artificial Intelligence

Abstract

Latent Diffusion Models (LDMs) achieve high-fidelity synthesis but suffer from latent space brittleness, causing discontinuous semantic jumps during editing. We introduce a Riemannian framework to diagnose this instability by analyzing the generative Jacobian, decomposing geometry into \textit{Local Scaling} (capacity) and \textit{Local Complexity} (curvature). Our study uncovers a \textbf{``Geometric Decoupling"}: while curvature in normal generation functionally encodes image detail, OOD generation exhibits a functional decoupling where extreme curvature is wasted on unstable semantic boundaries rather than perceptible details. This geometric misallocation identifies ``Geometric Hotspots" as the structural root of instability, providing a robust intrinsic metric for diagnosing generative reliability.

Keywords

Cite

@article{arxiv.2604.18804,
  title  = {Geometric Decoupling: Diagnosing the Structural Instability of Latent},
  author = {Yuanbang Liang and Zhengwen Chen and Yu-Kun Lai},
  journal= {arXiv preprint arXiv:2604.18804},
  year   = {2026}
}
R2 v1 2026-07-01T12:27:08.804Z