Personalized Image Generation
Utilizes single or multiple images that contain the same subject or style, along with text prompt, to generate images that contain that subject as well as match the textual description. Includes finetuning-based methods (e.g. DreamBooth, Textual Inversion) as well as encoder-based methods (e.g. E4T, ELITE, and IP-Adapter, etc.).
Papers
No papers found.
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | DreamBooth LoRA SDXL v1.0 | Overall (CP * PF) | 0.52 | — | Unverified |
| 2 | IP-Adapter ViT-G SDXL v1.0 | Overall (CP * PF) | 0.38 | — | Unverified |
| 3 | Emu2 SDXL v1.0 | Overall (CP * PF) | 0.36 | — | Unverified |
| 4 | DreamBooth SD v1.5 | Overall (CP * PF) | 0.36 | — | Unverified |
| 5 | IP-Adapter-Plus ViT-H SDXL v1.0 | Overall (CP * PF) | 0.34 | — | Unverified |
| 6 | BLIP-Diffusion SD v1.5 | Overall (CP * PF) | 0.27 | — | Unverified |
| 7 | Textual Inversion SD v1.5 | Overall (CP * PF) | 0.24 | — | Unverified |