Levy Tran Portraits (read description below if you want to know more about)

I wanted to experiment a more "structured" prompt to see what AI understands with some keywords.

After several tries i ended up with this mental scheme (keep in mind i tend to avoid traditions and manuals at some point in life i like to do stuff my way). Consider this as my vision of a reproducible scheme of prompt: style and aesthetic sometime seem to be interchangeable and they tend influence objects and mood on the pictures (like clothes and location). The subject, especially with proper names seems to work better when in " ".

The most influence on the picture i found comes from the author (you just write By "name of the author here"). Of course the LoRA used have to be tuned more and more and more until you find a balance and if the picture never seems to look how you have it in your mind it is time to consider different base model and adding or removing LoRA.

If you want to know how i think about slides... I suggest you to avoid going over 0.65. Usually things goes nuts over that value. But anyway sometime with details and peculiar effects maybe it is exactly what you want. To test a specific slide influence just leave everything else at 0.50 or 0.65 and pump this specific slide you want to test up to 2 or 1.75 and go down each time. Usually when you reach 1.25 the picture starts to become "understandable". But hey, this is just my way.