Who is Willow? Do you recognise her? You don't. She's a creation of Stable Diffusion and is here for your use in your prompts.I recommend you use ADetailer to show it at its best.Any buzz tips or ratings appreciated!Willow was me trying to get something to work, and mostly failing. When generating <a target="_blank" rel="ugc" href="https://github.com/BelieveDiffusion/tutorials/blob/main/consistent_character_embedding/README.md">images from Stable Diffusion to create my source images</a> (rather than using the FaceApp method) the characters I end up with seem flatter than the source images, and often missing key details in most engines. I've talked about black characters tending to become white.With Willow, I wanted to create a sort of cute bombshell character - Sandra Dee from Grease, Betty from the Archie books. 1960s good girl vibe. I tried a lot of methods on the dataset, and basically lost the blonde hair and blue eyes on almost all the methods, and some of the character's pop. She's obviously beautiful, but she wasn't the 100% knockout I was trying for.Trying all the methods did teach me quite a few things about what works and what doesn't. There's one more method I want to try with this data to see if it gets any better results. I got some which were more blonde more regularly, but they almost always wanted to make the character naked too.Which was best? It was a toss up between a slight variation of the one in the above tutorial and my standard method. You have to laugh.However, in the process, I did hit upon another idea, which worked perfectly - see my next model, <a rel="ugc" href="https://civarchive.com/models/178092/belinda-notexist">Belinda Notexist.</a>

Files with SHA256: 419af1b3bc6695fa6bc3557ecb7e441b06c32d7228954468c4b350e91019a618