How to use lama and powerpaint together?

[jryebread]

Model
lama and powerpaint-v1

Describe the bug
When I use both lama and powerpaint v1 inpainting, I don't see an option to use just lama in the UI, only powerpaint version of object removal, which is much slower and worse. How can I use lama when powerpaint is enabled? i think this is a bug.

System Info
Software version used

• iopaint: latest
• pytorch: 12.1
• CUDA:12.1

command to run it:
iopaint start --model=lama --device=cuda --model=Sanster/PowerPaint-V1-stable-diffusion-inpainting --device=cuda --host=0.0.0.0 --enable-realesrgan --realesrgan-model="RealESRGAN_x4plus" --realesrgan-device=cuda --enable-remove-bg --enable-interactive-seg --interactive-seg-model=vit_l --interactive-seg-device=cuda --disable-nsfw-checker

[Sanster]

Multiple models are not supported. --model=Sanster/PowerPaint-V1-stable-diffusion-inpainting is placed after --model=lama, so PowerPaint-V1-stable-diffusion-inpainting is used when the service starts. You can switch models in the settings on the web page.

[jryebread]

@Sanster appreciate your response and creating this cool project!

May i ask which inpainting model you have tried is best for real images? ive been testing them all but found none have worked well for my real image use case (trying to insert new people into existing images)