# GAE：为 3D 一致世界生成学习几何原生隐空间

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-09-21 08:00
- AIHOT 分数：41
- AIHOT 链接：https://aihot.news/items/cmudj6uwj0cokroggvsj6ivev
- 原文链接：https://arxiv.org/abs/2609.24981

## AI 摘要

研究者提出几何原生自编码器（GAE），将几何基础模型的特征重参数化为紧凑隐空间，其隐变量可联合解码为外观、深度、相机和点图。在固定生成器与训练协议的对照实验中，替换为 GAE 隐空间后，RealEstate10K 和 DL3DV 上的 FVD 分别下降 12.7% 和 23.1%，RealEstate10K 的相机轨迹误差减半。

## 正文

We present a compact geometry-native latent space as a shared foundation for perception and generation. Visual generators can produce photorealistic frames without preserving a consistent 3D scene. We argue that this is not only a modeling problem but also a representation problem: generators typically evolve appearance-centric latents, while perception models recover geometry in a semantically rich space that encodes cross-view structure. Rather than adding geometry as another output, we reparameterize a geometry foundation model's features into a compact latent space for generation.

We realize this shift with the geometry-native autoencoder (GAE), whose latent is jointly decodable to appearance, depth, cameras, and point maps. With this state, a standard conditional flow supports diverse generation tasks. In controlled comparisons that hold the generator and training protocol fixed, replacing the latent with GAE improves both visual quality and independently measured 3D coherence: FVD falls by 12.7% and 23.1% on RealEstate10K and DL3DV, and camera-trajectory error is halved on RealEstate10K. Together, these results show that the latent space is central to geometry-consistent generation and can serve as a shared interface between perception and generation.