Research map / Image generation / Text-to-image diffusion
EmotiCrafter: Text-to-Emotional-Image Generation based on Valence-Arousal Model
Shengqi Dang, Yi He, Long Ling, Ziqing Qian, Nanxuan Zhao, Nan Cao
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 V-A density weighting d(v,a) to the loss (ablation) | EmotiCrafter · Density weighting ablation | V-Error: 1.546 -> 1.510 | Improved |
| add V-A density weighting d(v,a) to the loss (ablation) | EmotiCrafter · Density weighting ablation | CLIPScore: 21.977 -> 23.067 | Improved |
| add V-A density weighting d(v,a) to the loss (ablation) | EmotiCrafter · Density weighting ablation | A-Error: 1.829 -> 1.828 | No clear effect |
| increase residual scale factor alpha (ablation) | EmotiCrafter · Emotional image generation | V/A-Error | Improved |
| increase residual scale factor alpha (ablation) | EmotiCrafter · Emotional image generation | CLIPScore | Hurt |
| rewrite prompts with emotion features via learned mapping network instead of GPT-4 | EmotiCrafter vs GPT-4+SDXL · Continuity | LPIPS-Continuous: 0.361 -> 0.220 | Improved |
| use emotion-embedding network instead of GPT-4 rewriting | EmotiCrafter vs GPT-4+SDXL · User study I | V-Error: 1.229 -> 0.692 | Improved |
| use emotion-embedding network instead of GPT-4 rewriting | EmotiCrafter vs GPT-4+SDXL · User study II | Emotion Smoothness: 3.195 -> 4.240 | Improved |
| embed V-A into prompt features instead of injecting into SDXL UNet (cross-attention) | EmotiCrafter vs Cross Attention · Emotional image generation | A-Error: 1.923 -> 1.828 | Improved |
| embed V-A into prompt features instead of injecting into UNet | EmotiCrafter vs Cross Attention/Time Embedding · Emotional image generation | CLIPScore: 26.566 -> 23.067 | Hurt |
Closest papers by meaning
- EmoGen: Emotional Image Content Generation with Text-to-Image Diffusion Models (CVPR 2024 · similarity 0.90)
- EmoVid: A Multimodal Emotion Video Dataset for Emotion-Centric Video Understanding and Generation (AAAI 2026 · similarity 0.87)
- EmoEdit: Evoking Emotions through Image Manipulation (CVPR 2025 · similarity 0.87)
- A Unified and Interpretable Emotion Representation and Expression Generation (CVPR 2024 · similarity 0.87)
- Customizing Visual Emotion Evaluation for MLLMs: An Open-vocabulary, Multifaceted, and Scalable Approach (ICLR 2026 · similarity 0.84)
- EmoVIT: Revolutionizing Emotion Insights with Visual Instruction Tuning (CVPR 2024 · 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.
