Sign In

Lookism | Mitsuki Soma

14
123
37
2
Verified:
SafeTensor
Type
LoRA
Stats
123
37
Reviews
Published
Mar 16, 2024
Base Model
SD 1.5
Training
Steps: 2,880
Epochs: 160
Usage Tips
Clip Skip: 2
Trigger Words
mitsuki soma, 1girl, solo, mole under eye, black eyes, red lips, ponytail, black hair, long hair, white shirt, white pants, white jacket, white suit, red necktie, hair ribbon,
mitsuki soma, 1girl, solo, (fox mask, bells), black eyes, red lips, ponytail, black hair, long hair, white shirt, white pants, white jacket, white suit, red necktie, hair ribbon,
mitsuki soma, 1girl, solo, mole under eye, black eyes, black hair, long hair, school uniform, blue sailor collar, long sleeves, pleated skirt, blue skirt, sailor collar, red neckerchief,
Hash
AutoV2
38757D85B1

A LoRa for the Character "Mitsuki Soma" out of the Webnovel "Lookism"

It comes with both her outfits (White suit + school uniform) and both of them with and without her mask

Recommended use with a strength of 0.7 - 0.85 but should work with more or less too.

Like always i highly recommend using Highresfix for better details.

tested on different models. Works with anime and semi realistic ones.

Random rambling:

I am not 100% happy with the LoRa in its current state but its good enough to share. Might need a couple gens to get the mask right and the face can be a bit iffy when using the mask.

Tried my best with what was available of the character. I try to make my LoRas as close to the original characters looks as possible while still offering flexibility. I achieve this by manually editing tags and using picture software to edit training images. Sometimes i get too focused on small details that bother me even though they are not that important. This often results in retraining the same Lora with minor changes. This explains the high numbers in my LoRa files.

i would much appreciate some likes and comments if you like using my work as it shows me people appreciate the work i put in. Also dont be shy on sharing some pictures you all generated as i would like to see how my models handle other prompts and models.

I wish everyone a nice day.