mlsanityData to 7 Oct 2026

TaoAvatar: Real-Time Lifelike Full-Body Talking Avatars for Augmented Reality via 3D Gaussian Splatting

arXiv 2503.17032 ↗First posted 21 Mar 2025cs.CVcited by 11

Not on the map. Why?

Abstract (arXiv)

Realistic 3D full-body talking avatars hold great potential in AR, with applications ranging from e-commerce live streaming to holographic communication. Despite advances in 3D Gaussian Splatting (3DGS) for lifelike avatar creation, existing methods struggle with fine-grained control of facial expressions and body movements in full-body talking tasks. Additionally, they often lack sufficient details and cannot run in real-time on mobile devices. We present TaoAvatar, a high-fidelity, lightweight, 3DGS-based full-body talking avatar driven by various signals. Our approach starts by creating a personalized clothed human parametric template that binds Gaussians to represent appearances. We then pre-train a StyleUnet-based network to handle complex pose-dependent non-rigid deformation, which can capture high-frequency appearance details but is too resource-intensive for mobile devices. To overcome this, we "bake" the non-rigid deformations into a lightweight MLP-based network using a distillation technique and develop blend shapes to compensate for details. Extensive experiments show that TaoAvatar achieves state-of-the-art rendering quality while running in real-time across various devices, maintaining 90 FPS on high-definition stereo devices such as the Apple Vision Pro.

Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). Builds on 56 of its 73 references that are in the map. First posted on arXiv 21 Mar 2025.

Builds on

56 of 73 references.

References · 56

  1. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,029
  2. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,141
  3. 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,698
  4. 2024Qwen2 Technical Reportcited by 2,076
  5. 2019Expressive Body Capture: 3D Hands, Face, and Body from a Single Imagecited by 1,600
  6. 20234D Gaussian Splatting for Real-Time Dynamic Scene Renderingcited by 1,144
  7. 2023Dynamic 3D Gaussians: Tracking by Persistent Dynamic View Synthesiscited by 731
  8. 2022HumanNeRF: Free-viewpoint Rendering of Moving People from Monocular Videocited by 483
  9. 2021Animatable Neural Radiance Fields for Modeling Dynamic Human Bodiescited by 361
  10. 2022NeuS2: Fast Learning of Neural Implicit Surfaces for Multi-view Reconstructioncited by 311
  11. 2023GaussianShader: 3D Gaussian Splatting with Shading Functions for Reflective Surfacescited by 249
  12. 2019Multi-Garment Net: Learning to Dress 3D People from Imagescited by 244
38 more
  1. 2018Video Based Reconstruction of 3D People Modelscited by 237
  2. 2021SNARF: Differentiable Forward Skinning for Animating Non-Rigid Neural Implicit Shapescited by 221
  3. 2023GS-IR: 3D Gaussian Splatting for Inverse Renderingcited by 220
  4. 20233DGS-Avatar: Animatable Avatars via Deformable 3D Gaussian Splattingcited by 195
  5. 2023GaussianAvatar: Towards Realistic Human Avatar Modeling from a Single Video via Animatable 3D Gaussianscited by 190
  6. 2024SplattingAvatar: Realistic Real-Time Human Avatars with Mesh-Embedded Gaussian Splattingcited by 171
  7. 2019Learning to Reconstruct People in Clothing from a Single RGB Cameracited by 168
  8. 2023GauHuman: Articulated Gaussian Splatting from Monocular Human Videoscited by 168
  9. 2023HumanRF: High-Fidelity Neural Radiance Fields for Humans in Motioncited by 161
  10. 2022Instant Volumetric Head Avatarscited by 157
  11. 2023HUGS: Human Gaussian Splatscited by 154
  12. 2022SelfRecon: Self Reconstruction Your Digital Avatar from Monocular Videocited by 144
  13. 2022Paraformer: Fast and Accurate Parallel Transformer for Non-autoregressive End-to-End Speech Recognitioncited by 139
  14. 2022InstantAvatar: Learning Avatars from Monocular Video in 60 Secondscited by 138
  15. 2021Real-time Deep Dynamic Characterscited by 133
  16. 2023Relightable Gaussian Codec Avatarscited by 126
  17. 2023GART: Gaussian Articulated Template Modelscited by 121
  18. 2021HumanNeRF: Efficiently Generated Human Radiance Field from Sparse Inputscited by 112
  19. 2024Bilateral Reference for High-Resolution Dichotomous Image Segmentationcited by 112
  20. 2021Animatable Neural Radiance Fields from Monocular RGB Videoscited by 109
  21. 2023FlashAvatar: High-fidelity Head Avatar with Efficient Gaussian Embeddingcited by 107
  22. 2023Human Gaussian Splatting: Real-time Rendering of Animatable Avatarscited by 105
  23. 2023DNA-Rendering: A Diverse Neural Actor Repository for High-Fidelity Human-centric Renderingcited by 103
  24. 2023AvatarReX: Real-time Expressive Full-body Avatarscited by 83
  25. 2022HOOD: Hierarchical Graphs for Generalized Modelling of Clothing Dynamicscited by 77
  26. 2023X-Avatar: Expressive Human Avatarscited by 66
  27. 2020MNN: A Universal and Efficient Inference Enginecited by 64
  28. 2023GAvatar: Animatable 3D Gaussian Avatars with Implicit Mesh Learningcited by 63
  29. 2023HeadGaS: Real-Time Animatable Head Avatars via 3D Gaussian Splattingcited by 60
  30. 2022Fast-SNARF: A Fast Deformer for Articulated Neural Fieldscited by 55
  31. 2023StyleAvatar: Real-time Photo-realistic Portrait Avatar from a Single Videocited by 52
  32. 2024PhysAvatar: Learning the Physics of Dressed 3D Avatars from Visual Observationscited by 51
  33. 2024GoMAvatar: Efficient Animatable Human Modeling from Monocular Video Using Gaussians-on-Meshcited by 51
  34. 20244D-DRESS: A 4D Dataset of Real-world Human Clothing with Semantic Annotationscited by 51
  35. 2023Learning Disentangled Avatars with Hybrid 3D Representationscited by 44
  36. 2024Generalizable Human Gaussians for Sparse View Synthesiscited by 40
  37. 2024MeshAvatar: Learning High-quality Triangular Human Avatars from Multi-view Videoscited by 34
  38. 2024UniTalker: Scaling up Audio-Driven 3D Facial Animation through A Unified Modelcited by 29

50 most cited of 56 All 56 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 3; 2026: 8.

11 papers.

  1. 2025A Survey on 3D Gaussian Splatting Applications: Segmentation, Editing, and Generationcited by 19
  2. 2025Bringing Your Portrait to 3D Presencecited by 3
  3. 2026OmniFit: Multi-modal 3D Body Fitting via Scale-agnostic Dense Landmark Predictioncited by 2
  4. 2026Mon3tr: Monocular 3D Telepresence with Pre-built Gaussian Avatars as Amortizationcited by 1
  5. 2026DeGS: A Scalable 3DGS Architecture via Decoupled Workload Parsing and Reorganizationcited by 1
  6. 2025PoseLLM: Enhancing Language-Guided Human Pose Estimation with MLP Alignmentcited by 0
  7. 2026Autoregressive Appearance Prediction for 3D Gaussian Avatarscited by 0
  8. 2026UNICA: A Unified Neural Framework for Controllable 3D Avatarscited by 0
  9. 2026LUNA: Learning Universal 3D Human Animation Beyond Skinningcited by 0
  10. 2026Astrolabe: Spherical-Map Guidance Across Diffusion Pipelines for Full-Body Capture from Unconstrained Imagescited by 0
  11. 2026AvatarDynamizer: From Static to Dynamic Human Avatars via Generative Dynamic Texturescited by 0