Pose Anything Easily with Img2Img in Stable Diffusion 2!

Stable Diffusion 2.0 now has a working Dreambooth version thanks to Huggingface Diffusers! There is even an updated script to convert the diffusers model into the original ckpt format, so you’re ready to go with whatever web UI you use. Even compared to the SD1.5 Inpainting Model, this version excels at Img2Img. Ideal if you want to make some personalised cards or presents - just in time for the end of the year! Get a head start with this initial release before everyone else plays catch-up over the next few weeks 😉 Update: The A111 extension “The Dreambooth training based on Shivam Shiaro’s repo“ has now been updated to support SD 2.0! Update: diffusers is no longer in the requirements file Install: 1.) conda create --name db-diffusers python= 2.) conda activate db-diffusers 3.) git clone 4.) cd diffusers 5.) pip install -e . 6.) cd examples/dreambooth 7.) pip install -U -r requirement