model does not work when process the image create by other gen model ,like flux

[kongzijian]

model does not work when process the image create by other gen model ,like flux.there is the result ,in which ,left is the gen result ,middle is depth map ,right one is ori image.

[wenquanlu]

Hi,
I think the depth map still alright, you can just replace the inpainting model with a better one, note that stable diffusion1.5 is pretty outdated and coarse. I would suggest try using a higher-res and better-finetuned inpainting model, it should work reasonable well as long as the depth map is correct

[kongzijian]

@wenquanlu thanks a lot , today i solve the problem by using mj7 ,by the way , i got a new problem when I use control_sd15_inpaint_depth_hand_fp16.safetensors which from lllyasviel with args.finetuned =True, the result as follow,in which ,left is the ori ,middle is depth map ,right one is gen result image.meanwhile I got correct result by using sd-v1-5-inpainting and control_v11f1p_sd15_depth together. I dont know why.