RelightAnyone：一种泛化的可重光照 3D 高斯头部模型
RelightAnyone: A Generalized Relightable 3D Gaussian Head Model
把平面光照下的 3DGS 头部 avatar 映射到基于物理的反射率参数，使没有 OLAT 采集的主体也能重光照；对做 3DGS avatar、逆渲染与单图重建的人，这是一个可复用的两阶段训练范式与光照编码对齐思路。
原文依据 learn a mapping from flat-lit 3DGS avatars to corresponding relightable Gaussian parameters for that avatar
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论3D 高斯泼溅（3DGS）已成为重建并渲染照片级真实感 3D 头部 avatar 的标准方法。一个主要挑战是如何对这些 avatar 进行重光照，使其匹配任意场景光照。为获得高质量重光照，现有方法要求受试者在复杂的时间复用光照下采集，例如一次一盏灯（OLAT）。我们提出一种新的泛化可重光照 3D 高斯头部模型，可对仅由单视图或多视图图像观测到的任意受试者进行重光照，而无需该受试者的 OLAT 数据。我们的核心思想是学习一个从平面光照 3DGS avatar 到该 avatar 对应可重光照高斯参数的映射。模型包含两个阶段：第一阶段在无 OLAT 光照的条件下建模平面光照 3DGS avatar；第二阶段学习到基于物理的反射率参数的映射，以实现高质量重光照。这种两阶段设计使我们能够在无 OLAT 光照的多个现有多视图数据集上训练第一阶段，从而保证跨受试者泛化，其中我们学习数据集特定的光照编码以进行自监督光照对齐。随后，第二阶段可以在一个规模显著更小的、在 OLAT 光照下采集的受试者数据集上训练。二者结合使我们的方法具备良好泛化能力，能够对第一阶段的任意受试者进行重光照，效果如同我们在 OLAT 光照下采集了他们。此外，我们还能仅用单张图像拟合未见过的受试者，从而在数字 avatar 的新视角合成与重光照方面支持多种应用。
英文原文摘要 1583 字符
3D Gaussian Splatting (3DGS) has become a standard approach to reconstruct and render photorealistic 3D head avatars. A major challenge is to relight the avatars to match any scene illumination. For high quality relighting, existing methods require subjects to be captured under complex time-multiplexed illumination, such as one-light-at-a-time (OLAT). We propose a new generalized relightable 3D Gaussian head model that can relight any subject observed in a single- or multi-view images without requiring OLAT data for that subject. Our core idea is to learn a mapping from flat-lit 3DGS avatars to corresponding relightable Gaussian parameters for that avatar. Our model consists of two stages: a first stage that models flat-lit 3DGS avatars without OLAT lighting, and a second stage that learns the mapping to physically-based reflectance parameters for high-quality relighting. This two-stage design allows us to train the first stage across diverse existing multi-view datasets without OLAT lighting ensuring cross-subject generalization, where we learn a dataset-specific lighting code for self-supervised lighting alignment. Subsequently, the second stage can be trained on a significantly smaller dataset of subjects captured under OLAT illumination. Together, this allows our method to generalize well and relight any subject from the first stage as if we had captured them under OLAT lighting. Furthermore, we can fit our model to unseen subjects from as little as a single image, allowing several applications in novel view synthesis and relighting for digital avatars.