High Quality Dreambooth (JP's repo / 24gb-Vram) trained checkpoint of Elon Musk using recently taken HD photographs from news/media sites.

There are a couple LORAs of Musk on Civitai, but imho the likeness achieved by this checkpoint will out perform any Lora model. Providing the full 2gb checkpoint (vs. an extracted / lossy LoRA file) will give you the best quality. If a CKPT for one likeness isn't practical/useful for your workflow then go check out the LoRA versions - please don't comment in regards to the fornat - this is for those w resources who want top quality. It's fairly easy to extract a Lora from checkpoints using various A1111 extensions (For example"Supermerger" is the one I use). If you do make a Lora from this checkpoint please keep it for personal use and don't circulated it in an extracted (lossy) format.

Use Hi-Res fix with a denoise strength in the .2 range for best results. For optimal results use with the Controlnet extension (w/ a model like Openpose or Canny cutting off before the final step)

Most of the examples were generated w/ controlnet so you won't be able to replicate them (w/o the image I used to initiate the generation). Also, I renamed the file prior to uploading but all of the samples were generated w/ this model (even if the info shows a different name).