- Object Addition
- Object Replacement
- Object Removal
- Background Change
- Style Change
- Texture Change
- Action Change
| Rank | Model | Publication | Dataset | LMM Score | |
|---|---|---|---|---|---|
| Mean↑ | Standard Deviation ↓ | ||||
| 1 | LEDITS++ | LEDITS++: Limitless Image Editing using Text-to-Image Models (CVPR 2024) | EditEval_v1 | 8.65 | 1.29 |
| 2 | ProxEdit | ProxEdit: Improving Tuning-Free Real Image Editing With Proximal Guidance (WACV 2024) | EditEval_v1 | 7.75 | 1.26 |
| 3 | InstructDiffusion | InstructDiffusion: A generalist modeling interface for vision tasks (CVPR 2024) | EditEval_v1 | 7.48 | 1.68 |
| 4 | AIDI | Effective real image editing with accelerated iterative diffusion inversion (ICCV 2023) | EditEval_v1 | 6.72 | 1.53 |
| 5 | Inst-Inpaint | Inst-Inpaint: Instructing to Remove Objects with Diffusion Models (arXiv 2023) | EditEval_v1 | 6.10 | 1.54 |
| Rank | Model | Publication | Dataset | LMM Score | |
|---|---|---|---|---|---|
| Mean↑ | Standard Deviation ↓ | ||||
| 1 | Null-Text Inversion | Null-text inversion for editing real images using guided diffusion models (CVPR 2023) | EditEval_v1 | 7.28 | 1.17 |
| 2 | LEDITS++ | LEDITS++: Limitless Image Editing using Text-to-Image Models (CVPR 2024) | EditEval_v1 | 6.91 | 0.97 |
| 4 | InstructPix2Pix | InstructPix2Pix: Learning to follow image editing instructions (CVPR 2023) | EditEval_v1 | 6.51 | 2.49 |
| 3 | ProxEdit | ProxEdit: Improving Tuning-Free Real Image Editing With Proximal Guidance (WACV 2024) | EditEval_v1 | 6.35 | 0.78 |
| 6 | DDS | Delta denoising score (ICCV 2023) | EditEval_v1 | 6.17 | 0.81 |
| 7 | AIDI | Effective real image editing with accelerated iterative diffusion inversion (ICCV 2023) | EditEval_v1 | 5.95 | 2.34 |
| 8 | EDICT | EDICT: Efficient Diffusion for Image Composition and Transformation (CVPR 2023) | EditEval_v1 | 5.01 | 1.94 |
| 5 | SINE | Sine: Single image editing with text-to-image diffusion models (CVPR 2023) | EditEval_v1 | 3.02 | 1.83 |
| Rank | Model | Publication | Dataset | LMM Score | |
|---|---|---|---|---|---|
| Mean↑ | Standard Deviation ↓ | ||||
| 1 | InstructPix2Pix | InstructPix2Pix: Learning to follow image editing instructions (CVPR 2023) | EditEval_v1 | 8.21 | 0.40 |
| 2 | InstructDiffusion | InstructDiffusion: A generalist modeling interface for vision tasks (CVPR 2024) | EditEval_v1 | 7.41 | 0.66 |
| 3 | Imagic | Imagic: Text-based real image editing with diffusion models (CVPR 2023) | EditEval_v1 | 7.26 | 1.25 |
| 4 | Null-Text Inversion | Null-text inversion for editing real images using guided diffusion models (CVPR 2023) | EditEval_v1 | 7.16 | 0.90 |
| 5 | DDPM Inversion | An edit friendly ddpm noise space: Inversion and manipulations (CVPR 2024) | EditEval_v1 | 7.08 | 0.81 |
| 6 | AIDI | Effective real image editing with accelerated iterative diffusion inversion (ICCV 2023) | EditEval_v1 | 6.99 | 1.07 |
| 8 | LEDITS++ | LEDITS++: Limitless Image Editing using Text-to-Image Models (CVPR 2024) | EditEval_v1 | 6.86 | 1.20 |
| 7 | ProxEdit | ProxEdit: Improving Tuning-Free Real Image Editing With Proximal Guidance (WACV 2024) | EditEval_v1 | 6.80 | 1.07 |
| Rank | Model | Publication | Dataset | LMM Score | |
|---|---|---|---|---|---|
| Mean↑ | Standard Deviation ↓ | ||||
| 1 | Imagic | Imagic: Text-based real image editing with diffusion models (CVPR 2023) | EditEval_v1 | 6.97 | 0.80 |
| 2 | Forgedit | Forgedit: Text Guided Image Editing via Learning and Forgetting (arXiv 2023) | EditEval_v1 | 6.38 | 0.87 |
| 3 | MasaCtrl | MasaCtrl: Tuning-Free Mutual Self-Attention Control for Consistent Image Synthesis and Editing (ICCV 2023) | EditEval_v1 | 6.55 | 1.16 |
| 4 | TIC | Tuning-Free Inversion-Enhanced Control for Consistent Image Editing (arXiv 2023) | EditEval_v1 | 6.20 | 1.70 |