arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.03319cs.GRcs.AIcs.LG
收藏

CaricatureGS: 通过高斯曲率夸大3D高斯点云人脸

CaricatureGS: Exaggerating 3D Gaussian Splatting Faces With Gaussian Curvature

Technion – Israel Institute of Technology(技术学院–以色列理工学院)

Eldad Matmon, Amit Bracha, Noam Rotstein, Ron Kimmel

AI总结:

CaricatureGS通过高斯曲率夸大技术,结合3D高斯点云和局部仿射变换,实现可控且逼真的卡通化虚拟形象生成。

摘要:

本文介绍了一种逼真且可控的3D卡通化框架。我们从基于内在高斯曲率的表面夸大技术开始,当与纹理结合时,倾向于产生过度平滑的渲染效果。为了解决这个问题,我们采用最近已被证明能产生真实自由视角虚拟形象的3D高斯点云(3DGS)。给定一个多视角序列,我们提取FLAME网格,求解曲率加权泊松方程,并获得其夸张形式。然而,直接变形高斯点云会产生较差的结果,需要通过局部仿射变换将每个帧扭曲到其夸张的2D表示来合成伪地面真实卡通图像。然后我们设计了一种训练方案,交替使用真实和合成监督,使单个高斯集合能够表示自然和夸张的虚拟形象。该方案提高了保真度,支持局部编辑,并允许对卡通强度进行连续控制。为了实现实时变形,引入了原表面和夸张表面之间的高效插值。我们进一步分析并证明其与闭式解的偏差是有界的。在定量和定性评估中,我们的结果优于先前工作,交付了逼真且几何可控的卡通虚拟形象。

英文摘要:

A photorealistic and controllable 3D caricaturization framework for faces is introduced. We start with an intrinsic Gaussian curvature-based surface exaggeration technique, which, when coupled with texture, tends to produce over-smoothed renders. To address this, we resort to 3D Gaussian Splatting (3DGS), which has recently been shown to produce realistic free-viewpoint avatars. Given a multiview sequence, we extract a FLAME mesh, solve a curvature-weighted Poisson equation, and obtain its exaggerated form. However, directly deforming the Gaussians yields poor results, necessitating the synthesis of pseudo-ground-truth caricature images by warping each frame to its exaggerated 2D representation using local affine transformations. We then devise a training scheme that alternates real and synthesized supervision, enabling a single Gaussian collection to represent both natural and exaggerated avatars. This scheme improves fidelity, supports local edits, and allows continuous control over the intensity of the caricature. In order to achieve real-time deformations, an efficient interpolation between the original and exaggerated surfaces is introduced. We further analyze and show that it has a bounded deviation from closed-form solutions. In both quantitative and qualitative evaluations, our results outperform prior work, delivering photorealistic, geometry-controlled caricature avatars.

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。

↑