Original Description:
Melkor Mancin Style.
I tried to capture his style as much as possible, also focusing on details. I really love Melkor’s works and followed him for several years. I hope this model would also help more readers to approach to his incredible work!
Changelog:
V4 [Pony XL]
Trained with .
New version trained with enriched dataset, better tags and training parameters.
(same dataset and parameters of V4 [Illustrious XL])
Notice: this was an experiment that seems to have worked better in Illustrious, perhaps the training parameters for Pony were not so good.
It seems to be weaker than V3 but gives more consistent results with the basic version of Pony and despite being weaker used even at high levels it is more compatible with the other Lora so it might still be useful in some scenarios, I personally still prefer V3.
V4 [Illustrious XL]
Trained with Illustrious-XL .
New version trained with enriched dataset, better tags and training parameters.
V3 [Pony XL]
New version trained with enriched dataset, better tags and training parameters.
New concepts:
"pulling out"
"bikini_tanlines"
"brac3s" (this does't work very well :( )
fluids: given more importance to fluids as in the original style
V2 [Illustrious XL]
Trained with Illustrious-XL with the same dataset of Melkor Mancin Style [Pony XL] V2
V2 [Pony XL]:
Trained with with better dataset and training parameters. The model is now able to trigger the characters:
Chloe Currey
Sidney Swanson
Janice Hanson
Pamela Anniculopoloboah
Kelly Swanson
Timothy Currey
Use the trigger words of the model, if some traits are not triggered add them to the generation, es.: "dark-skinned female" for Pam.
V1 [Pony XL]:
First release. Trained with on subset of pictures that better represents what I personally prefer of Melkor's style.