Quick question about model used

Open surya-narayanan opened this issue 1 year ago • 1 comments

Hi,

I was just trying to see what model you use - it looks like you have a link that points to stable diffusion, but I was a bit confused since stable diffusion generally does text-to-image without a mask. Is it a specific variant of sd/sdxl that does inpainting like this one?

Regards, Surya

Apr 04 '24 19:04 surya-narayanan

It's using an older version of Stable Diffusion that had init_image and mask as inputs.

Code: https://github.com/replicate/inpainter/blob/92274fe8c460d93873f12e454975ac985b3c84f3/pages/api/predictions/index.js#L16-L18

Model version: https://replicate.com/stability-ai/stable-diffusion/versions/be04660a5b93ef2aff61e3668dedb4cbeb14941e62a3fd5998364a32d613e35e

We should probably update it to use SDXL which is faster, higher quality, and has support for init images and masks. See https://replicate.com/stability-ai/sdxl

If you wanna take a shot at updating that in a pull request, I'd be happy to review it and ship it. :)

Apr 04 '24 21:04 zeke