A well trained LoRA of musician Ed Sheeran. Trained w/ Dreambooth then a HQ LoRA was extracted from the trained model using Kohya scripts. Trigger with, "Edsheeran Person" (Yes, you should include person) Trained on SDXL base 1.0 so likeness will come out best on that. It'll "work" with custom models, but depending on how modified they are from the base the likeness may or may not come through. **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 the likeness seems "weak").

Files with SHA256: 910934e061b9a8e18dc476a88ab51e3e88fc23594ab965347c47b71487323559