Research map / Image generation / Text-to-image diffusion
Semantic Line Diffusion: Character-Consistent Line Art from text-annotated Storyboards
Seo-Yeon Choi, Kyungsu Lee
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 multi-reference conditioning (ablation) | PanelDiff · Ablation, 50-cut sequences | MS-SSIM: 0.861 -> 0.903 | Improved |
| add multi-reference conditioning (ablation) | PanelDiff · Ablation, 50-cut sequences | identity drift: 0.301 -> 0.214 | Improved |
| use multiple references instead of single reference (ablation) | PanelDiff · Ablation, 50-cut sequences | MS-SSIM: 0.878 -> 0.903 | Improved |
| add memory bank (ablation) | PanelDiff · Ablation, 50-cut sequences | identity drift: 0.332 -> 0.214 | Improved |
| add memory bank (ablation) | PanelDiff · Ablation, 50-cut sequences | MS-SSIM: 0.842 -> 0.903 | Improved |
| use full memory instead of recent-only memory (ablation) | PanelDiff · Ablation, 50-cut sequences | identity drift: 0.259 -> 0.214 | Improved |
| add similarity interpreter (ablation) | PanelDiff · Ablation, 50-cut sequences | identity drift: 0.317 -> 0.214 | Improved |
| add similarity interpreter (ablation) | PanelDiff · Ablation, 50-cut sequences | MS-SSIM: 0.851 -> 0.903 | Improved |
| increase static references K from 1 to 8 (ablation) | PanelDiff · Reference count sweep | MS-SSIM: 0.78 -> 0.88 | Improved |
| increase static references K from 1 to 8 (ablation) | PanelDiff · Reference count sweep | identity drift: 0.32 -> 0.26 | Improved |
| increase conti perturbation noise from 0.0 to 1.0 | PanelDiff · Robustness | MS-SSIM: 0.90 -> 0.79 | Hurt |
| increase conti perturbation noise from 0.0 to 1.0 | Baseline average · Robustness | MS-SSIM: 0.82 -> 0.48 | Hurt |
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.
- DiffSketcher — DiffSketcher: Text Guided Vector Sketch Synthesis through Latent Diffusion Models (NeurIPS 2023) · better on 6 of 7 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
- One-Shot Reference-based Structure-Aware Image to Sketch Synthesis (AAAI 2025 · similarity 0.87)
- Storynizor: Consistent Story Generation via Inter-Frame Synchronized and Shuffled ID Injection (AAAI 2025 · similarity 0.85)
- DiffSensei: Bridging Multi-Modal LLMs and Diffusion Models for Customized Manga Generation (CVPR 2025 · similarity 0.85)
- StoryWeaver: A Unified World Model for Knowledge-Enhanced Story Character Customization (AAAI 2025 · similarity 0.85)
- Storybooth: Training-Free Multi-Subject Consistency for Improved Visual Storytelling (ICLR 2025 · similarity 0.85)
- SpotActor: Training-Free Layout-Controlled Consistent Image Generation (AAAI 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.
