Research map / Video / Video generation
STARCaster: Spatio-Temporal AutoRegressive Video Diffusion for Identity- and View-Aware Talking Portraits
Foivos Paraperas Papantoniou, Stathis Galanakis, Rolandos Alexandros Potamias, Bernhard Kainz, Stefanos Zafeiriou
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 lip-reading loss (ablation) | STARCaster · TH-1KH | LSE-C: 5.140 -> 5.493 | Improved |
| add lip-reading loss (ablation) | STARCaster · TH-1KH | LSE-D: 9.016 -> 8.724 | Improved |
| add self-forcing instead of teacher forcing (ablation) | STARCaster · TH-1KH | Pose Std (x10^-2): 4.386 -> 4.896 | Improved |
| add self-forcing instead of teacher forcing (ablation) | STARCaster · TH-1KH | FVD: 197.28 -> 185.24 | Improved |
| add self-forcing instead of teacher forcing (ablation) | STARCaster · Hallo3 | Pose Std (x10^-2): 4.161 -> 4.760 | Improved |
| use progressive training stages instead of a single autoregressive stage (ablation) | STARCaster · Hallo3 | FVD: 152.98 -> 145.35 | Improved |
| use progressive training instead of a single stage (ablation) | STARCaster · Hallo3 | LSE-C: 6.233 -> 6.292 | No clear effect |
| drive generation by ID embedding only without portrait reference | STARCaster · TH-1KH | LSE-C: 5.493 -> 5.627 | Improved |
| use lightweight SD-UNet backbone instead of 5B DiT (Hallo3) | audio-driven animation · 5-second clip inference | inference time (min): 21.4 -> 2.2 | Improved |
| use implicit video-based view synthesis instead of 3D inversion (Real3DPortrait) | 3D-aware talking portraits · NeRSemble | PSNR: 10.289 -> 14.219 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Hallo3 — Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer (CVPR 2025) · better on 6 of 6 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
- Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Video Diffusion Transformer (CVPR 2025 · similarity 0.88)
- GoHD: Gaze-oriented and Highly Disentangled Portrait Animation with Rhythmic Poses and Realistic Expressions (AAAI 2025 · similarity 0.86)
- RealPortrait: Realistic Portrait Animation with Diffusion Transformers (AAAI 2025 · similarity 0.86)
- IM-Portrait: Learning 3D-aware Video Diffusion for Photorealistic Talking Heads from Monocular VideosC (CVPR 2025 · similarity 0.86)
- StarGen: A Spatiotemporal Autoregression Framework with Video Diffusion Model for Scalable and Controllable Scene Generation (CVPR 2025 · similarity 0.86)
- X-NeMo: Expressive Neural Motion Reenactment via Disentangled Latent Attention (ICLR 2025 · similarity 0.86)
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.
