Research map / Image generation / Text-to-image diffusion
CharaConsist: Fine-Grained Consistent Character Generation
Mengyu Wang, Henghui Ding, Jianing Peng, Yao Zhao, Yunpeng Chen, Yunchao Wei
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 point-tracking attention (ablation) | FLUX.1 consistent generation · Background maintaining | ID Sim: 0.476 -> 0.572 | Improved |
| add adaptive token merge on top of point-tracking attention (ablation) | CharaConsist · Background maintaining | ID Sim: 0.572 -> 0.645 | Improved |
| add foreground-background masking for background consistency (ablation) | CharaConsist · Background maintaining | CLIP-I-bg: 0.899 -> 0.916 | Improved |
| add foreground-background masking (ablation) | CharaConsist · Background maintaining | CLIP-I-fg: 0.883 -> 0.883 | No clear effect |
| add all three components to FLUX.1 (ablation) | FLUX.1 · Background maintaining | CLIP-I: 0.874 -> 0.910 | Improved |
| add the proposed modules (image quality cost) (ablation) | FLUX.1 · Image quality | IQS: 0.989 -> 0.985 | Hurt |
| add CharaConsist to base model (consistency increment) (ablation) | FLUX.1 · Background maintaining | ID Sim: +.171 | Improved |
| use CharaConsist instead of ConsiStory | FLUX.1 vs SDXL · Background maintaining | CLIP-I-bg: 0.895 -> 0.916 | Improved |
| use CharaConsist instead of StoryDiffusion | FLUX.1 vs RealVisXL4.0 · Background switching | ID Sim: 0.563 -> 0.587 | Improved |
| use CharaConsist (training-free) instead of PhotoMaker (trained) | FLUX.1 · Background maintaining | ID Sim: 0.730 -> 0.647 | Hurt |
| use CharaConsist instead of StoryDiffusion | FLUX.1 vs RealVisXL4.0 · Background maintaining | CLIP-T: 0.370 -> 0.356 | Hurt |
| use CharaConsist instead of StoryDiffusion | FLUX.1 · Background switching | CLIP-I: 0.853 -> 0.840 | Hurt |
Later papers that compared with this method
1 later accepted paper reports CharaConsist as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- The Devil is in Attention Sharing: Improving Complex Non-rigid Image Editing Faithfulness via Attention Synergy (CVPR 2026 · better on 6 of 6 shared results)
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 4 of 6 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
- SpotActor: Training-Free Layout-Controlled Consistent Image Generation (AAAI 2025 · similarity 0.86)
- Storybooth: Training-Free Multi-Subject Consistency for Improved Visual Storytelling (ICLR 2025 · similarity 0.86)
- CCM: Real-Time Controllable Visual Content Creation Using Text-to-Image Consistency Models (ICML 2024 · similarity 0.85)
- Focus-N-Fix: Region-Aware Fine-Tuning for Text-to-Image Generation (CVPR 2025 · similarity 0.85)
- DreamText: High Fidelity Scene Text Synthesis (CVPR 2025 · similarity 0.85)
- MMIG-Bench: Towards Comprehensive and Explainable Evaluation of Multi-Modal Image Generation Models (NeurIPS 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.
