Single-frame Image Consistency on MSBench Single-subject v1
0.872CLIP Image ConsistencyFluxKontext
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FluxKontextCategory=In-context, Time (s)=32.31, VRAM (GB)=22.992026.02 | 0.872 | 0.859 | 0.342 | |
| NanoBananaCategory=In-context2026.02 | 0.857 | 0.898 | 0.362 | |
| StoryTailorCategory=Ours, Time (s)=7.02, VRAM (GB)=20.012026.02 | 0.849 | 0.811 | 0.431 | |
| Qwen-EditCategory=In-context2026.02 | 0.841 | 0.902 | 0.373 | |
| MS-DiffusionCategory=Adapters & MLLM, Time (s)=5.36, VRAM (GB)=19.472026.02 | 0.824 | 0.848 | 0.327 | |
| SSR-EncoderCategory=Adapters & MLLM, Time (s)=7.49, VRAM (GB)=18.822026.02 | 0.813 | 0.799 | 0.331 | |
| IP-AdapterCategory=Adapters & MLLM, Time (s)=4.82, VRAM (GB)=19.392026.02 | 0.809 | 0.616 | 0.283 | |
| DreamBoothCategory=Fine-tuning, Time (s)=4.25, VRAM (GB)=13.122026.02 | 0.802 | 0.589 | 0.309 | |
| LoRACategory=Fine-tuning, Time (s)=3.85, VRAM (GB)=15.822026.02 | 0.795 | 0.591 | 0.27 | |
| λ-ECLIPSECategory=Adapters & MLLM, Time (s)=8.31, VRAM (GB)=14.212026.02 | 0.792 | 0.806 | 0.308 | |
| Textual InversionCategory=Fine-tuning, Time (s)=3.51, VRAM (GB)=13.732026.02 | 0.772 | 0.574 | 0.264 |