Research map / 3D vision / Face and shape
Beyond Fixed Topologies: Unregistered Training and Comprehensive Evaluation Metrics for 3D Talking Heads
Federico Nocentini, Thomas Besnier, Claudio Ferrari, Sylvain Arguillère, Mohamed Daoudi, Stefano Berretti
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add mask and velocity losses (ablation) | ScanTalk (md) · VOCAset | FDD: 2.10 -> 1.47 | Improved |
| add mask and velocity losses (ablation) | ScanTalk (md) · VOCAset | DTW: 1.39 -> 1.14 | Improved |
| add mask and velocity losses (ablation) | ScanTalk (md) · VOCAset | LVE: 6.38 -> 7.03 | Hurt |
| add mask, velocity and cosine losses (ablation) | ScanTalk (md) · VOCAset | FDD: 2.10 -> 1.29 | Improved |
| add mask, velocity and cosine losses (ablation) | ScanTalk (md) · VOCAset | delta_Cd: 0.60 -> 0.52 | Improved |
| add mask, velocity and cosine losses (ablation) | ScanTalk (md) · BIWI6 | LVE: 4.04 -> 4.44 | Hurt |
| train on multiple datasets instead of single dataset (ablation) | ScanTalk · Multiface | FDD: 64.43 -> 32.20 | Improved |
| train on multiple datasets instead of single dataset (ablation) | ScanTalk · VOCAset | LVE: 3.01 -> 6.38 | Hurt |
| use Chamfer loss instead of MSE on registered data (ablation) | ScanTalk · VOCAset | LVE: 6.38 -> 6.57 | Hurt |
| train on remeshed data instead of registered (Chamfer loss) (ablation) | ScanTalk · VOCAset | FDD: 2.31 -> 8.22 | Hurt |
| add Gaussian noise to remeshed data with holes (ablation) | ScanTalk · VOCAset | FDD: 9.94 -> 39.0 | Hurt |
| random test-time rotation of +-30 degrees around X axis (ablation) | ScanTalk · VOCAset | LVE increase (%): 14.9% increase (X) vs 4.8 (Y) and 5.0 (Z) | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- VOCA — VOCA: Visual Odometry with Codec Awareness (ECCV 2026) · better on 4 of 4 shared results
- CodeTalker — CodeTalker: Speech-Driven 3D Facial Animation With Discrete Motion Prior (CVPR 2023) · better on 1 of 4 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
- Perceptually Accurate 3D Talking Head Generation: New Definitions, Speech-Mesh Representation, and Evaluation Metrics (CVPR 2025 · similarity 0.92)
- Seeing Speech: Learning Visible Articulatory Dynamics for Speech-Driven 3D Facial Animation (NeurIPS 2026 · similarity 0.90)
- Probabilistic Speech-Driven 3D Facial Motion Synthesis: New Benchmarks Methods and Applications (CVPR 2024 · similarity 0.90)
- FaceTalk: Audio-Driven Motion Diffusion for Neural Parametric Head Models (CVPR 2024 · similarity 0.89)
- PointTalk: Audio-Driven Dynamic Lip Point Cloud for 3D Gaussian-based Talking Head Synthesis (AAAI 2025 · similarity 0.88)
- AnyTalk: Multi-modal Driven Multi-domain Talking Head Generation (AAAI 2025 · similarity 0.88)
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.
