Research map / Video / Video generation
Phantom: Subject-Consistent Video Generation via Cross-Modal Alignment
Lijie Liu, Tianxiang Ma, Bingchuan Li, Zhuowei Chen, Jiawei Liu, Gen Li, Siyu Zhou, Qian He, Xinglong Wu
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 VAE features alongside CLIP (ablation) | Phantom-IP · Single-subject S2V | CLIP-I: 0.693 -> 0.714 | Improved |
| add VAE features alongside CLIP (ablation) | Phantom-IP · Single-subject S2V | Aes score: 62.03 -> 64.32 | Improved |
| add CLIP features alongside VAE (ablation) | Phantom-IP · Single-subject S2V | CLIP-I: 0.512 -> 0.714 | Improved |
| add CLIP features alongside VAE (ablation) | Phantom-IP · Single-subject S2V | DINO-I: 0.302 -> 0.523 | Improved |
| add CLIP features alongside VAE (ablation) | Phantom-IP · Single-subject S2V | Aes score: 48.82 -> 64.32 | Improved |
| add VAE features alongside CLIP (ablation) | Phantom-IP · Single-subject S2V | ViCLIP-T: 23.63 -> 23.40 | Hurt |
| add text-image alignment in training (ablation) | Phantom · Multi-subject generation | success rate: 65% -> 95% | Improved |
| Phantom-ID vs Hailuo-ID | Hailuo-ID · Face ID video generation | FaceSim-Arc: 0.542 -> 0.581 | Improved |
| Phantom-IP vs Kling1.6 | Kling1.6 · Single-subject S2V | CLIP-I: 0.732 vs 0.714 | Hurt |
| Phantom-IP vs Kling1.6 on prompt following | Kling1.6 · Single-subject S2V | ViCLIP-T: 21.62 -> 23.41 | Improved |
Closest papers by meaning
- Phantom-Data: Towards a General Subject-Consistent Video Generation Dataset (ICLR 2026 · similarity 0.91)
- PolyVivid: Vivid Multi-Subject Video Generation with Cross-Modal Interaction and Enhancement (NeurIPS 2025 · similarity 0.88)
- Neuro-Symbolic Evaluation of Text-to-Video Models using Formal Verification (CVPR 2025 · similarity 0.87)
- Large-Scale Text-to-Image Model with Inpainting is a Zero-Shot Subject-Driven Image Generator (CVPR 2025 · similarity 0.87)
- Video-Bench: Human-Aligned Video Generation Benchmark (CVPR 2025 · similarity 0.86)
- OpenS2V-Nexus: A Detailed Benchmark and Million-Scale Dataset for Subject-to-Video Generation (NeurIPS 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.
