Research map / Video / Video generation
DreamingComics: A Story Visualization Pipeline via Subject and Layout Customized Generation using Video Models
Patrick Kwon, Chen Chen
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 RegionalRoPE (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CIDS cross: 38.7 -> 58.5 | Improved |
| add RegionalRoPE (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CSD cross: 41.0 -> 52.9 | Improved |
| add masked condition loss (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CIDS cross: 56.6 -> 58.5 | Improved |
| add masked condition loss (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CSD cross: 50.5 -> 52.9 | Improved |
| use target timestamp t=3 instead of t=1 (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CIDS cross: 55.3 -> 58.5 | Improved |
| use target timestamp t=3 instead of t=9 (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CIDS cross: 54.7 -> 58.5 | Improved |
| use target timestamp t=3 instead of t=5 (ablation) | Dream-Illustrator · ViStoryBench (ablation, 6K steps single-subject) | CIDS cross: 55.7 -> 58.5 | Improved |
| replace GPT-4 layout generator with fine-tuned Qwen2.5-VL | layout generator · Comic layout generation | Page coverage ratio: 29.25 -> 79.14 | Improved |
| replace GPT-4 layout generator with fine-tuned Qwen2.5-VL | layout generator · Comic layout generation | Panel ordering: 66.67 -> 99.00 | Improved |
| use video-DiT based Dream-Illustrator instead of DreamO | image customization · ViStoryBench | CIDS cross: 51.6 -> 66.6 | Improved |
| use video-DiT based Dream-Illustrator instead of image generator baselines | image customization · ViStoryBench | Aesthetics score: 4.54 vs 5.81 (Eligen) | Hurt |
| use Dream-Illustrator instead of DreamO | image customization · DreamBench | CLIP-T: 32.59 vs 32.96 | No clear effect |
Closest papers by meaning
- StoryGPT-V: Large Language Models as Consistent Story Visualizers (CVPR 2025 · similarity 0.88)
- DiffSensei: Bridging Multi-Modal LLMs and Diffusion Models for Customized Manga Generation (CVPR 2025 · similarity 0.87)
- ViStoryBench: Comprehensive Benchmark Suite for Story Visualization (CVPR 2026 · similarity 0.87)
- SpotActor: Training-Free Layout-Controlled Consistent Image Generation (AAAI 2025 · similarity 0.87)
- StoryWeaver: A Unified World Model for Knowledge-Enhanced Story Character Customization (AAAI 2025 · similarity 0.87)
- LogiStory: A Logic-Aware Framework for Multi-Image Story Visualization (ICLR 2026 · 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.
