High Quality Dreambooth (JP's 24gb repo) trained LoRA of influencer Logan Paul created using high-resolution images. Trained with Dreambooth (24gb) and then the LoRA was extracted using the A1111 extension "<a rel="ugc" href="https://github.com/hako-mikan/sd-webui-supermerger">Supermerger</a>"Trigger with "Loganpaul Person"(yes, put person also). Use "Hi-Res fix" with the settings you'd normally use for best results. I recommend Eular_a. Use prompt weighting and/or Controlnet to get maximum flexibility. Use the original Stable Diffusion models (v1.4, v1.5, etc) for best likeness results. ==Note on the Samples:Some same images were generated using Control Net (particularly the 16x9 film style images). I don't think Civitai notes if Controlnet was used but you may not be able to reproduce some of the images due to that. (I often use canny w/ a random movie still for a % of the early steps to guide the image to do something interesting). If there's interest in the full checkpoint (2gb but no loss) I can add that as an additional version down the road. 34images / 5000 steps trained locally.

Files with SHA256: cf8bc1e7c1b90a97668f9ed623ce258d0debfececd5bbbdc44b2fda710c129ca