This my first Style LoRA trained with the creations of "Double Deck". It's also provide some characters from his work, Even if this can be a pretty niche style, I'm hoping you may find what you where searching for.
Even with a very small dataset, it do work with realistic and furry style.
The keyword is 'double deck, monster girl' but you can use specific character like :
- 'mushroom girl, plant girl', 'anthro girl, frog girl', 'anthro girl, raccoon girl' for animal/plant characters
- 'robot girl', 'android', 'tesla girl' and 'assaultron' to generate robot
- 'silent hill' plus 'pyramid head' or 'bubble head nurse', which are more chaotic during generation
- 'white and black' for the original style, or 'colored picture' to force painted variant of the style
Some arguments may exist but aren't usefull enough or may break the generation, like 'multiple breasts' or others...
For the Strength of this LoRA, it seem to work OK on 1, but It work best by lowering a bit values if there is artefact during generation.
Description
strength can go from 0.8 up to 1.2
New dataset with Fallout content
More human on human content
tried to add closed shot for better details
added experimental insect monster
FAQ
Comments (8)
Works great, it's probably the hardest to do but is there a chance to add the silent hill mannequin?
In fact there is few pictures with the mannequin, but I didn't use a specific keyword to train it... Just "headless, silent hill, mannequin, multiples breasts, multiples legs, multiples limbs, unusual anatomy" but it just won't get out. Perhaps I should try to train a SD15 lora with a specific trigger word and see if it work before redoing a Pony Lora
would lora's work with it?
@Lazius Yeah I figured it would be really troublesome even with enough sample images
@mcmanus I'm on my way for a experimental v3.1 with higher epoch and steps with specific trigger tag like mannequin_silent_hill for the silent_hill character. I'm at 75% of training and it finally start to look humanoid and not just a bunch of limbs, but I fear it may overtrain the rest of the lora character.
For reference, the V3 is 13.860 step, the expected steps for next version could be up to 23100
@Lazius We believe in you!
@mcmanus you should try the v3.1 for pony, for the next time I'm documenting myself on DoRA and experiment with a SD15 variant to see how it work
Double deck is the best, this Lora is gold