Michele Morrone (SDXL)

Michele Morrone is an Italian actor, singer, and model. Born on October 3, 1990, in Bitonto, Italy, he gained widespread recognition for his role as Massimo Torricelli in the 2020 Polish romantic drama film "365 Days" (365 Dni). The film, based on a novel by Blanka Lipińska, generated significant attention for its and became popular on streaming platforms.

VERSION 2

Trained on over 100 more images of Michele Morrone, better quality trained images, added trigger words, better flexibility and likeness, better captioning. I recommend using it at a strength of 0.9 but 1 also works well. 0.8 if you're feeling the need for more flexibility. As for the trigger word it's m1chele but when I try that it comes up with a woman so I just put "m1chele a man". "Beard" is also a main trigger word and I recommend using it. I've only used this lora on Juggernaut V7 but it should work on other models too. Negative prompts help but aren't needed, I recommend using the ones I used in the sample images.

VERSION 1

This is my first lora, I like how it turned out. It may not be the best at laying down or some other poses because I couldn't find many of those ones. This lora is trained on 58 images of Michele Morrone. All images in the preview have Adetailer. No hires fix or anything. I trained a Sd 1.5 one but it didn't turn out how I wanted. I've only used Juggernaut but I'm sure it works good with other models too. No trigger word just put "photo of a man". You can also add the word "beard" to your positive prompt. If you experience any problems let me know.

UPDATE

This actually works best at a strength of around 0.8. I recommend not using negative prompts since that's what gave me the best results but it depends on different things like the model you use or what you're aiming to do. This lora can also do him wearing glasses(some images had them in the training data). Probably works better with hi-res but I can't try it because my computer doesn't allow me to lol. Big thanks to diogod https://civitai.com/user/diogod for pointing this out. May artifacts because of the images I used to train the lora but shouldn't be too impactful as long as you use the 0.8 strength. If they are you could try putting jpeg artifacts in negative prompt. My next Loras shouldn't have this problem.