SDXL LoRA of Harrison Ford's character "Indiana Jones". This was created using hi-resolution images from the first 3 movies (young/middle aged) and a small amount of filler late 80s HF photos. The model was trained using Dreambooth(~fine tuning), and the LoRA was extracted from the trained model using Kohya scripts.Trigger with "Indianajones Person" - (Yes you should include "person")Trained on SDXL base 1.0 so likeness will come out best on that, but the model should work with models trained on SDXL also. Play with weighting for flexibility. (Re: change the LoRA weight in the &lt;&gt; code to .8, .7, .6 etc if you needed to get more creative control, or try raising it to 1.2, 1.4, etc if Indy's seems "weak"). I ran a number of trials on this one and it was hard enough to get a checkpoint I was satisfied with let alone a LoRA extraction (which does incur a bit of loss which may or may not hurt the likeness. Hope it works for those who grab it.

Files with SHA256: 1dddd2ff08a2b3d1250c5fe8284d6c37db834b8709d2d4e4ce348b4f44d66185