Personalized Image Generation On Dreambench
Metrics
Concept Preservation (CP)
Overall (CP * PF)
Prompt Following (PF)
Results
Performance results of various models on this benchmark
Comparison Table
| Model Name | Concept Preservation (CP) | Overall (CP * PF) | Prompt Following (PF) |
|---|---|---|---|
| dreambooth-fine-tuning-text-to-image | 0.494 | 0.356 | 0.721 |
| dreambooth-fine-tuning-text-to-image | 0.598 | 0.517 | 0.865 |
| ip-adapter-text-compatible-image-prompt | 0.833 | 0.344 | 0.413 |
| generative-multimodal-models-are-in-context | 0.528 | 0.364 | 0.690 |
| an-image-is-worth-one-word-personalizing-text | 0.378 | 0.236 | 0.624 |
| ip-adapter-text-compatible-image-prompt | 0.593 | 0.380 | 0.640 |
| blip-diffusion-pre-trained-subject-1 | 0.547 | 0.271 | 0.495 |