Research map / Vision-language and multimodal / CLIP and zero-shot
StoryTailor:A Zero-Shot Pipeline for Action-Rich Multi-Subject Visual Narratives
Jinghao Hu, Yuhe Zhang, Guohua Geng, Kang Li, Han Zhang
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 Action-Boost SVR (ablation) | StoryTailor · Multi-subject narratives | CLIP-T: 0.364 -> 0.406 | Improved |
| add Selective Forgetting Cache (ablation) | StoryTailor · Multi-subject narratives | CLIP-T: 0.341 -> 0.406 | Improved |
| add Gaussian-Centered Attention (ablation) | StoryTailor · Multi-subject narratives | CLIP-T: 0.319 -> 0.406 | Improved |
| add Gaussian-Centered Attention (ablation) | StoryTailor · Multi-subject narratives | CLIP-I: 0.684 -> 0.671 | Hurt |
| add Gaussian-Centered Attention (ablation) | StoryTailor · Multi-subject narratives | Subject confusion: Y -> N | Improved |
| add Action-Boost SVR (ablation) | StoryTailor · Multi-subject narratives | CLIP-I: 0.636 -> 0.671 | Improved |
| use two-stage anisotropic dynamic Gaussian (GCA) instead of binary box mask (ablation) | GCA · Multi-subject action scenes | CLIP-T: 0.285 -> 0.416 | Improved |
| make Gaussian radii dynamic rather than static (ablation) | GCA · Multi-subject action scenes | CLIP-T: 0.354 -> 0.416 | Improved |
| use two-stage instead of single-stage Gaussian (ablation) | GCA · Multi-subject action scenes | Subject confusion: Y -> N | Improved |
| use Gaussian diffusion split instead of GCA (ablation) | GCA · Multi-subject action scenes | Latency (s): 7.57 -> 31.32 | Hurt |
| raise AB-SVR energy threshold tau from 0.80 to 0.92 or 0.98 (ablation) | AB-SVR · Qualitative sweep, dog and dog+cat | identity and pose quality | Hurt |
| lower AB-SVR energy threshold tau to 0.50 or 0.56 (ablation) | AB-SVR · Qualitative sweep, dog and dog+cat | action fidelity | Hurt |
Closest papers by meaning
- Storybooth: Training-Free Multi-Subject Consistency for Improved Visual Storytelling (ICLR 2025 · similarity 0.87)
- Storynizor: Consistent Story Generation via Inter-Frame Synchronized and Shuffled ID Injection (AAAI 2025 · similarity 0.86)
- EntityBench: Towards Entity-Consistent Long-Range Multi-Shot Video Generation (NeurIPS 2026 · similarity 0.86)
- ContextualStory: Consistent Visual Storytelling with Spatially-Enhanced and Storyline Context (AAAI 2025 · similarity 0.85)
- Advancing Narrative Long Video Generation via Training-Free Identity-Aware Memory (NeurIPS 2026 · similarity 0.85)
- StoryGPT-V: Large Language Models as Consistent Story Visualizers (CVPR 2025 · similarity 0.85)
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.
