GLIGEN vs BoxDiff

[andupotorac]

Hi. In the paper you're mentioning that BoxDiff can work as a plug and play with GLIGEN. But I want to ask if you can provide more details. Don't the two projects do the same thing?

[Sierkinhane]

Hi, there is no layout constraint on attention in GLIGEN's inference. You can consider GLIGEN as a finetuned Diffusion model, then our training-free boxdiff can also guide its diffusion process, making it better in layout2image.

[andupotorac]

Thanks. So this is mainly useful for when someone doesn't add the constraints to GLIGEN manually? For example as a default preset that automatically sets the areas to speed up the UX?