Research map / Image generation / Text-to-image diffusion
Say Cheese! Detail-Preserving Portrait Collection Generation via Natural Language Edits
Zelong Sun, Jiahui Wu, Ying Ba, Dong Jing, Zhiwu Lu
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 supervised fine-tuning on CHEESE (ablation) | IP-Adapter (zero-shot) · CHEESE test set | Qwen-PF: 0.428 -> 0.683 | Improved |
| add supervised fine-tuning on CHEESE (ablation) | IP-Adapter (zero-shot) · CHEESE test set | CLIP-T: 0.406 -> 0.421 | Improved |
| add supervised fine-tuning on CHEESE (ablation) | IP-Adapter (zero-shot) · CHEESE test set | DINO-I: 0.824 -> 0.728 | Hurt |
| add ConsistencyNet (ablation) | + SFT · CHEESE test set | Qwen-DP: 0.783 -> 0.832 | Improved |
| add ConsistencyNet (ablation) | + SFT · CHEESE test set | Qwen-PF: 0.683 -> 0.673 | Hurt |
| add Fusion IP-Adapter (ablation) | + ConNet · CHEESE test set | Qwen-PF: 0.673 -> 0.723 | Improved |
| add alignment loss (ablation) | + Fusion IP-A · CHEESE test set | Qwen-PF: 0.723 -> 0.777 | Improved |
| add alignment loss (ablation) | + Fusion IP-A · CHEESE test set | DINO-I: 0.732 -> 0.753 | Improved |
| add teacher forcing (ablation) | + Align Loss · CHEESE test set | Qwen-DP: 0.836 -> 0.855 | Improved |
| add teacher forcing (ablation) | + Align Loss · CHEESE test set | Qwen-PF: 0.777 -> 0.793 | Improved |
| add teacher forcing (ablation) | + Align Loss · CHEESE test set | CLIP-T: 0.426 -> 0.436 | Improved |
| validate annotations with CLIP ViT-G/14 instead of ViT-B/32 in inversion-based validation (ablation) | CHEESE annotation pipeline · modification text quality (qualitative) | text quality | 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.
- EasyRef — EasyRef: Omni-Generalized Group Image Reference for Diffusion Models via Multimodal LLM (ICML 2025) · 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
- MagiCapture: High-Resolution Multi-Concept Portrait Customization (AAAI 2024 · similarity 0.86)
- PortraitBooth: A Versatile Portrait Model for Fast Identity-preserved Personalization (CVPR 2024 · similarity 0.85)
- DreamSalon: A Staged Diffusion Framework for Preserving Identity-Context in Editable Face Generation (CVPR 2024 · similarity 0.85)
- EchoShot: Multi-Shot Portrait Video Generation (NeurIPS 2025 · similarity 0.84)
- Text2Relight: Creative Portrait Relighting with Text Guidance (AAAI 2025 · similarity 0.83)
- HiFi-Portrait: Zero-shot Identity-preserved Portrait Generation with High-fidelity Multi-face Fusion (CVPR 2025 · similarity 0.83)
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.
