Type | |
Stats | 262 237 1 |
Reviews | (27) |
Published | Nov 24, 2023 |
Base Model | |
Training | Steps: 4,000 Epochs: 6 |
Usage Tips | Clip Skip: 2 Strength: 0.6 |
Trigger Words | 1girl long hair red hair thighhighs boots tube top cleavage gloves green eyes skirt +1 more |
Hash | AutoV2 0FF51D7F53 |
READ ME!
This is a LoRA that I only had three images initially to work with, so it suffers from some "model inbreeding" and may require ControlNet or lower weights to pose dynamically.
Recommended Weight: 0.8
to 0.4
depending on how accurate you want the outfit to be versus pose/background flexibility.
Example Prompt: 1girl, mature female, solo, sitting, full body, red hair, ponytail, long hair, tight (white gloves) with gold trim, green eyes, cleavage, (white tube top) with white sides, strapless, white boots with gold trim, black thighhighs with white (stars), bare shoulders, belt, midriff, bare arms, sleeveless, pleated skirt, tight clothes,<lora:StS-Emma-CravingQuest:0.6>, <lora:more_details:0.4>
ALSO I recommend putting elbow gloves
, detached sleeves
, and jacket
in the negative prompt.
This was a unique challenge to create a LoRA from so few references, and all things considered I think it came out well. A couple deficiencies of note are:
1.) Emma has an asymmetrical pattern on one of here thigh-highs, I was not able to train for this due to almost non-existent reference images. Instead I focused on black thigh-highs.
2.) Emma's choker is not shaped quite right, but it does have the star(s)
3.) Emma's boots often lack the red tassle, but it does put a nice start there most times.
4.) As with most SD 1.5 image generations, generating her holding a handgun/weapon is possible, but the weapon will likely look incoherent. There was little I could do about that with so few reference images.
I may update this model later with better version, but quite frankly doing this was a ton of work. Nevertheless I hope you enjoy.