follow me on Instagram, Youtube and my website.
Suggested Weight: 0.6-1.2 (this is where the embedding shines imo, you can dial it up or down as much as you need, depending on how well it's mixing with the other subject)
portrait photo of (man:1.0) as (q4n0n5h4m4n:0.9), horns, facepaint
portrait photo of (man:1.0) as (female q4n0n5h4m4n:0.9), horns, facepaint
portrait photo of (man:1.0) as (q4n0n5h4m4n:0.9), horns:0.3, facepaint:0.6
To give someone horns, you usually need to add horns
to the prompt, which depending on the subject and rest of the prompt might need to be weighted up or down (too high will turn them into a goat, too low won't render horns at all).
Also facepaint
can be most specific like patriotic facepaint
trained on ~12 images
Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005
Batch Size: 1
Gradient Steps: 1