Region-based text-to-image generation on COCO 2017 (val)
70.93FIDMultiDiffusion (Ref.)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MultiDiffusion (Ref.)Sampler=DDIM [52], Model=SD1.5, Resolution=512 x 512, Time (s)=14.12024.03 | 70.93 | 16.24 | 24.09 | 27.55 | — | |
| MultiDiffusionSampler=DDIM [52], Backbone=SD1.5, Resolution=512x512, Time (s)=14.12024.03 | 70.93 | 16.24 | 24.09 | 27.55 | — | |
| MultiDiffusion (Ref.)Sampler=DDIM [52], Model=SDXL (1024 x 1024), Time (s)=50.62024.03 | 73.77 | 16.31 | 24.16 | 28.11 | — | |
| SemanticDrawBackbone=SD3 (1024 x 1024), Sampler=FlashFlowMatch [8]2024.03 | 79.2 | 17.41 | 23.59 | 27.83 | 3.2 | |
| SemanticDrawSampler=EulerDiscrete [18], Model=SDXL (1024 x 1024), Time (s)=3.62024.03 | 84.27 | 15.04 | 24.19 | 24.22 | — | |
| SemanticDrawSampler=LCM [33], Backbone=SD1.5, Resolution=512x512, Time (s)=1.32024.03 | 93.93 | 14.12 | 24.14 | 24 | — | |
| SemanticDrawSampler=Hyper-SD [44], Model=SD1.5, Resolution=512 x 512, Time (s)=1.32024.03 | 98.6 | 14.9 | 24.48 | 23.31 | — | |
| MultiDiffusion (Ref.)Backbone=SD3 (1024 x 1024), Sampler=FlowMatch [10]2024.03 | 166.42 | 8.517 | 20.66 | 16.39 | 46.3 | |
| MultiDiffusion (MD)Sampler=Hyper-SD [44], Model=SD1.5, Resolution=512 x 512, Time (s)=1.72024.03 | 168.34 | 10.12 | 20.08 | 15.9 | — | |
| MultiDiffusion (MD)Backbone=SD3 (1024 x 1024), Sampler=FlashFlowMatch [8]2024.03 | 209.36 | 5.347 | 19.83 | 14.48 | 4 | |
| MultiDiffusionSampler=LCM [33], Backbone=SD1.5, Resolution=512x512, Time (s)=1.72024.03 | 270.55 | 2.653 | 22.53 | 19.63 | — | |
| MultiDiffusion (MD)Sampler=EulerDiscrete [18], Model=SDXL (1024 x 1024), Time (s)=4.32024.03 | 572.95 | 1.328 | 21.02 | 17.36 | — |