Nickel || Dungeon Munchies

This is the first time I've trained with such a small dataset (4 images), I've retrained this model 5 times with different parameters and now I think it's somewhat decent enough to share. I'm also working on Grill. Used kohya_ss gui.

Here are some parameters and helpful keywords:

• Checkpoint: AbyssOrangeMix3

• Trigger Word: nickel

• Weight:

• 0 - 0.4 ( Little to no similarity )

• 0.5 - 0.75 ( More similarity with room to control )

• 0.75 - 1 ( Harder to control, details/angle/colors harder to modify, worse quality )

Could benefit from other models that help with color/saturation especially at higher weights.

Keywords:

yellow eyes, silver hair, headset

Default Outfit: vest, bowtie, short sleeves

( Sometimes hair can end up long, prompting hair color or adding short hair can help. )

To produce similar results I recommend EasyNegative, this negative embedding was used in the sample images. I'm also using Clip Skip 2 and hires fix.