GTAvatar: Bridging Gaussian Splatting and Texture Mapping for Relightable and Editable Gaussian Avatars

Edit & Relight PBR Material Support Animate Input video Figure 1: GTAvatar broadens applications of monocular Gaussian Splatting head avatars beyond reenactment and relighting, enabling inter- active editing of textures for precise control of intrinsic appearance, while preserving training efficiency, rendering speed and visual fidelity. Abstract Recent advancements in Gaussian Splatting have enabled increasingly accurate reconstruction of photorealistic head avatars, opening the door to numerous applications in visual effects, videoconferencing, and virtual reality. This, however, comes with the lack of intuitive editability offered by traditional triangle mesh–based methods. In contrast, we propose a method that com- bines the accuracy and fidelity of 2D Gaussian Splatting with the intuitiveness of UV texture mapping. By embedding each canonical Gaussian primitive’s local frame into a patch in the UV space of a template mesh in a computationally efficient man- ner, we reconstruct continuous editable material head textures from a single monocular video on a conventional UV domain. Furthermore, we leverage an efficient physically based reflectance model to enable relighting and editing of these intrinsic material maps. Through extensive comparisons with state-of-the-art methods, we demonstrate the accuracy of our reconstruc- tions, the quality of our relighting results, and the ability to provide intuitive controls for modifying an avatar’s appearance and geometry via texture mapping without additional optimization.

1. Introduction Gaussian Splatting based avatars have revolutionized the capture and rendering of digital humans by delivering an unprecedented level of photorealism with real-time rendering capability, enabling new possibilities for reanimating content from video inputs. How- ever, this realism comes with a trade-off: unlike traditional texture- map-based modeling, Gaussian avatars offer little flexibility for in- tuitive appearance editing. This gap becomes critical in practice. In arXiv:2512.09162v1 [cs.CV] 9 Dec 2025 2 of 16 K. Baert et al. / GTAvatar: Bridging Gaussian Splatting and Texture Mapping for Relightable and Editable Gaussian Avatars film and visual effects, artists routinely adjust the smallest details of a face - smoothing skin to create a flawless appearance, removing distracting high-frequency features, or sculpting age and character through wrinkles, scars and bruises. In gaming and virtual produc- tion, creators seek the same level of control to personalize avatars with tattoos, makeup, or stylized patterns using artist-friendly tools that operate directly on material texture maps. Without such editing capabilities, even the most realistic avatar remains a fixed reproduc- tion, rather than a medium for creative expression. This limitation stems from the fact that each Gaussian splat car- ries its own color or material properties in isolation, without the shared structure that texture maps naturally provide. As a result, the surface lacks the local coherence needed to treat it as a continuous canvas, making it difficult — if not impossible — to apply edits consistently, whether to a small region of the face or to its over- all appearance. Conversely, while mesh-based 3D inverse rendering methods [MHS∗22, BZH∗23, BBC∗24] inherit a coherent defacto UV-domain fit for artist-friendly edits, they can suffer from topo- logical rigidity when representing high-frequency details and are hampered in their reconstruction capacity due to the limitations of differentiable rasterization when handling complex or translucent geometries (cf. Table 1). Furthermore, naively embedding Gaus- sians in UV domain leads to discontinuous texture maps that are hard to edit successfully, as witnessed in our results (e.g. Figure 10) and also in [ZWL∗25]. The authors of FATE [ZWL∗25] proposed a second U-Net neural baking stage to alleviate this issue in the context of non-relightable avatars. Our goal is to reconstruct a head avatar from a single monocu- lar RGB video, that can be rendered in real time, relit under arbi- trary environment maps, animated with new poses and expressions, and directly edited through texture mapping, yet without the added postprocessing complexity of training a two-stage model or baking lighting information in texture directly. Following seminal work [QKS∗24, LKB∗25a], we adopt FLAME-anchored [LBB∗17] Gaussian splats. Our key observation is that, unlike vanilla 3DGS [KKLD23a], which operates directly in screen space, the 2DGS variant [HYC∗24] defines a local splat co- ordinate frame and computes ray–splat intersections to query ker- nels. We leverage this property to map ray–splat intersections into the UV domain via an approximate orthographic proj

…(Full text truncated)…