List your company
ResearchFormerly WizWand

Multimodal-to-Image Generation on MS-COCO 30k (val)

8.53FID (30k)MULTIFUSION
7.902412.138716.37520.6113May 24, 2023
Updated 8mo ago

Evaluation results

MethodLinks
MULTIFUSIONGuidance Scale=4, Prompt Type=Multimodal (Caption + Reference Image), Image Size=256x256, Diffusion Steps=1002023.058.53
MULTIFUSIONGuidance Scale=2, Prompt Type=Multimodal (Caption + Reference Image), Image Size=256x256, Diffusion Steps=1002023.058.61
MULTIFUSIONGuidance Scale=6, Prompt Type=Multimodal (Caption + Reference Image), Image Size=256x256, Diffusion Steps=1002023.0510.29
MULTIFUSIONGuidance Scale=8, Prompt Type=Multimodal (Caption + Reference Image), Image Size=256x256, Diffusion Steps=1002023.0511.5
MULTIFUSIONGuidance Scale=1, Prompt Type=Multimodal (Caption + Reference Image), Image Size=256x256, Diffusion Steps=1002023.0524.22