arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.14844cs.GRcs.AI
收藏

CAG-Avatar: 基于交叉注意力引导的高保真头部重建高斯姿态

CAG-Avatar: Cross-Attention Guided Gaussian Avatars for High-Fidelity Head Reconstruction

Dept. of Control Science and Engineering, University of Shanghai for Science and Technology(控制科学与工程系,上海科技大学);School of Psychology and Neuroscience, University of Glasgow(心理学与神经科学学院,格拉斯哥大学)

Zhe Chang, Haodong Jin, Yan Song, Hui Yu

AI总结:

CAG-Avatar通过交叉注意力引导的条件自适应机制,提升高保真头部重建的精度与实时渲染性能。

摘要:

创建高保真的实时可驾驶3D头部姿态是数字动画中的核心挑战。虽然3D高斯喷射(3D-GS)提供了前所未有的渲染速度和质量,但当前的动画技术往往依赖于一种“一刀切”的全局调节方法,其中所有高斯原语都由单一表达代码统一驱动。这种简单的方法无法揭示不同面部区域的动态差异,例如可变形的皮肤与刚性的牙齿,导致显著的模糊和失真伪影。我们引入了条件自适应高斯姿态(CAG-Avatar),一种解决这一关键限制的框架。其核心是一个基于交叉注意力的条件自适应融合模块。该机制使每个3D高斯原语充当查询,根据其标准位置自适应地从全局表达代码中提取相关驱动信号。这种“量身定制”的条件策略显著增强了对细粒度、局部动态的建模。我们的实验确认了重建保真度的显著提升,特别是在牙齿等具有挑战性的区域,同时保持了实时渲染性能。

英文摘要:

Creating high-fidelity, real-time drivable 3D head avatars is a core challenge in digital animation. While 3D Gaussian Splashing (3D-GS) offers unprecedented rendering speed and quality, current animation techniques often rely on a "one-size-fits-all" global tuning approach, where all Gaussian primitives are uniformly driven by a single expression code. This simplistic approach fails to unravel the distinct dynamics of different facial regions, such as deformable skin versus rigid teeth, leading to significant blurring and distortion artifacts. We introduce Conditionally-Adaptive Gaussian Avatars (CAG-Avatar), a framework that resolves this key limitation. At its core is a Conditionally Adaptive Fusion Module built on cross-attention. This mechanism empowers each 3D Gaussian to act as a query, adaptively extracting relevant driving signals from the global expression code based on its canonical position. This "tailor-made" conditioning strategy drastically enhances the modeling of fine-grained, localized dynamics. Our experiments confirm a significant improvement in reconstruction fidelity, particularly for challenging regions such as teeth, while preserving real-time rendering performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑