Stable Diffusion released a new doodle model (no current options for local installs)
Stable Diffusion released a new doodle model (no current options for local installs)
As the title says, Stability AI has released their new model called Stable Doodles, which lets you draw doodle then output an images based on it. Seems it is only limited to 14 styles at the moment and it is a combination of SDXL and T2I Adapters. Hopefully we will hear about a local install later but not entirely sure we will see it but due to it basically working via SDXL and T2I adapters, chances are the community will make something that is equal if not better in the coming weeks/months after the official release of SDXL.
How is it meaningfully different to the existing Scribble and Lineart controlnets that are already working in Automatic1111?
Functionally its working off the same mechanics. The only edge this would have is this is working off SDXL which is trained on 1024 x 1024 images. So you may see better details. As I made mention in my post there is a strong hope once we have SDXL, which is slated for public release in a few days we as a community will develop tools to reincorporate Scribble and Lineart controlnet seeing as those will likely break. Also chances are people are already working off of the leaked SDXL model to make such tools as we speak.