List your company
ResearchFormerly WizWand

Text-to-Image Generation on Multi-subject prompts (test)

0.3489CLIP I-TDivide&Bind
0.3427640.3443570.345950.347543Oct 2, 2025
Updated 7mo ago

Evaluation results

MethodLinks
Divide&BindBackbone=SD 3.5, λ=optimal2025.100.34890.23160.57420.639922.67791.34933.9373
Attend&ExciteBackbone=SD 3.5, λ=optimal2025.100.34840.23260.57520.640422.6951.35453.1714
FOCUSBackbone=SD 3.5, λ=optimal2025.100.34830.23440.57510.638522.74991.40034.2865
CONFORMBackbone=SD 3.5, λ=optimal2025.100.34810.23230.57730.642122.71881.36843.4336
BaseBackbone=SD 3.5, λ=optimal2025.100.34740.23090.57310.640222.6941.31750
Divide&BindBackbone=FLUX.1, λ=optimal2025.100.34530.22720.57220.63323.43951.29391.6352
BaseBackbone=FLUX.1, λ=optimal2025.100.34490.22710.57390.6323.42341.2970
FOCUSBackbone=FLUX.1, λ=optimal2025.100.34460.22680.57410.632623.42741.29131.9712
CONFORMBackbone=FLUX.1, λ=optimal2025.100.34360.22520.57260.632123.35741.24611.5114
Attend&ExciteBackbone=FLUX.1, λ=optimal2025.100.3430.22420.57160.630423.25491.24941.7595