Research map / Image generation / Text-to-image diffusion
Consistent Text-to-Image Generation via Scene De-Contextualization
Song Tang, Peihao Gong, Kunyu LI, Kai Guo, Boyu Wang, Mao Ye, Jianwei Dr. Zhang, Xiatian Zhu
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 |
|---|---|---|---|
| use soft learned estimation of the shared subspace instead of a hard correlation matrix (ablation) | SDeC · ConsiStory+ | DreamSim-F: 0.3351 -> 0.2589 | Improved |
| use absolute eigenvalue excursion rather than relative normalization (ablation) | SDeC · ConsiStory+ | DreamSim-F: 0.3576 -> 0.2589 | Improved |
| use soft learned estimation (ablation) | SDeC · ConsiStory+ | DreamSim-B: 0.4254 -> 0.3675 | Hurt |
| add SDeC prompt embedding editing to SDXL | SDXL · ConsiStory+ | DreamSim-F: 0.2778 -> 0.2589 | Improved |
| add SDeC prompt embedding editing to SDXL | SDXL · ConsiStory+ | DreamSim-B: 0.3861 -> 0.3675 | Hurt |
| combine SDeC with ConsiStory | ConsiStory · ConsiStory+ | DreamSim-F: 0.2729 -> 0.2542 | Improved |
| combine SDeC with ConsiStory | ConsiStory · ConsiStory+ | CLIP-I: 0.8604 -> 0.8744 | Improved |
| replace 1P1S prompt re-structuring (w/o IPCA) with SDeC embedding editing | 1Prompt1Story w/o IPCA · ConsiStory+ | DreamSim-F: 0.2682 -> 0.2589 | Improved |
| place scene prompt before ID prompt (reverse order) (ablation) | SDeC · ConsiStory+ | DreamSim-F: 0.2589 -> 0.2670 | Hurt |
| place scene prompt before ID prompt (reverse order) (ablation) | SDXL · ConsiStory+ | DreamSim-F: 0.2778 -> 0.2894 | Hurt |
| use SDeC instead of training-based methods | consistent T2I · ConsiStory+ user study | win rate (%): 42.67% vs 20.83% (ConsiStory) | 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.
- PhotoMaker — PhotoMaker: Customizing Realistic Human Photos via Stacked ID Embedding (CVPR 2024) · better on 1 of 1 shared result
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
- Localizing and Editing Knowledge In Text-to-Image Generative Models (ICLR 2024 · similarity 0.88)
- One-Prompt-One-Story: Free-Lunch Consistent Text-to-Image Generation Using a Single Prompt (ICLR 2025 · similarity 0.87)
- Continuous, Subject-Specific Attribute Control in T2I Models by Identifying Semantic Directions (CVPR 2025 · similarity 0.87)
- XVerse: Consistent Multi-Subject Control of Identity and Semantic Attributes via DiT Modulation (NeurIPS 2025 · similarity 0.87)
- Prompt-Free Diffusion: Taking "Text" out of Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.87)
- LLM Blueprint: Enabling Text-to-Image Generation with Complex and Detailed Prompts (ICLR 2024 · 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.
