AI Paper Atlas

Research map / 3D vision / Face and shape

Tavatar: Topology-Aware Gaussian Attribute Derivation for Animatable Human Avatars

Hailin Luo, Yifan Yang, Jiazhi Shu, Zixiong Huang, Qi Chen, Qing Du, Mingkui Tan

CVPR 2026 · 3D vision › Face and shape · approach: Gaussian and implicit head avatars

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
derive Gaussian scale and rotation analytically from mesh instead of optimizing themTavatar vs best baseline (IHuman) · X-AvatarNormal error: 2.056 -> 1.772Improved
derive Gaussian attributes analytically from meshTavatar vs best baseline (IHuman) · PeopleSnapshotNormal error: 2.055 -> 1.687Improved
derive Gaussian attributes analytically from meshTavatar vs best baseline (IHuman) · X-AvatarChamfer Distance: 0.132 -> 0.111Improved
analytic topology-aware binding vs free Gaussians (GART) on in-distribution rotationsTavatar · PeopleSnapshotPSNR: male-3: 30.21 (GART) vs 28.93; male-4: 27.55 vs 27.56Mixed
analytic topology-aware binding on OOD posesTavatar · X-AvatarPSNR: 00016: 28.86 -> 29.03; 00017: 25.24 -> 25.24Mixed
add Face Gaussians (analytically derived) (ablation)Tavatar · X-Avatar subject 00019PSNR: 26.89 -> 28.11Improved
add Vertex Gaussians (ablation)Tavatar · X-Avatar subject 00019PSNR: 25.67 -> 28.11Improved
add equilateral regularization (ablation)Tavatar · X-Avatar subject 00019PSNR: 27.83 -> 28.11Improved
add equilateral regularization (ablation)Tavatar · X-Avatar subject 00019Normal error: 1.834 -> 1.772Improved
add Vertex Gaussians (ablation)Tavatar · X-Avatar subject 00019Normal error: 2.687 -> 1.772Improved
add Face Gaussians (ablation)Tavatar · X-Avatar subject 00019Normal error: 2.143 -> 1.772Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy