Text-to-Image Generation on In-the-wild image color condition
50.9FIDIP-Adapter
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| IP-AdapterEvaluation Protocol=Baseline, Inference Cost=1.442025.03 | 50.9 | 22 | 10.1 | 13.9 | |
| Color-Aligned DiffusionEvaluation Protocol=Fine-tune, Conditioning Space=Latent, Inference Cost=1.032025.03 | 69.4 | 27.4 | 4.98 | 4.87 | |
| Stable DiffusionEvaluation Protocol=Pre-train, Inference Cost=1.002025.03 | 72.6 | 27.3 | 27.8 | 16 | |
| Style-AlignedEvaluation Protocol=Baseline, Inference Cost=5.692025.03 | 73.1 | 22.9 | 6.18 | 8.5 | |
| T2I-AdapterEvaluation Protocol=Baseline, Inference Cost=1.562025.03 | 77.1 | 24.9 | 8.34 | 7.01 | |
| Color-Aligned DiffusionEvaluation Protocol=Zero-shot, Conditioning Space=Latent, Inference Cost=3.522025.03 | 77.9 | 27.4 | 2.68 | 3.93 | |
| ControlNet-ColorEvaluation Protocol=Baseline, Inference Cost=1.782025.03 | 81 | 23.4 | 13.1 | 6.54 |