In your paper, you mention that the reference image is inpainted using SD-inpainting with the prompt “background.” However, when I use "background" as the inpainting prompt, the fooocus model often generates unwanted objects. Could you provide more details on the reference inpainting process?


In your paper, you mention that the reference image is inpainted using SD-inpainting with the prompt “background.” However, when I use "background" as the inpainting prompt, the fooocus model often generates unwanted objects. Could you provide more details on the reference inpainting process?