
This is my first LoRA with Illustrious and I'm really surprised how accurate it can be, anyway it may not be quite right, but it looks good to me.
step: 30+| strength: 0.9 (1-0.85) | cfg scale: 6 (7-) | clip skip: 2 | sampler: Euler A
For correct anatomy according to the movie, use:
-
tbatb-jiromaru, 1boy, chubby,
unibrow, solo,To be more precise:
(gnome:0.5),negative prompt:
gnome, fur skin,-
red scarf, yellow jacket, white t-shirt, red sash, white baggy shorts, brown shoes
-
red scarf, orange vest, white sweater, red sash, white baggy shorts, brown shoes
-
-
tbatb-jteen, 1boy, fat,
unibrow, solo,To be more precise: young,
-
red scarf, white jacket, white t-shirt, brown sash, beige baggy shorts, brown shoes
-
optional changes: chubby -> fat
negative prompt: (tail),
Recommended
prompt: masterpiece, best quality, amazing quality, colorful, vibrant, very aesthetic,
negative prompt: bad quality, worst quality, worst detail, normal quality, sketch, censor, tail, deformed pupils, deformed eyes, bad eyes, tits, breasts, feminization, bar censor, censored, mosaic censoring, from side,
negative prompt: bar censor, censored, mosaic censoring, bad quality, worst quality, worst detail, normal quality, sketch, censor, (tail), deformed pupils, deformed eyes, bad eyes, tits, breasts, feminization, (two-tone skin), pixelated, dark skin, deformed penis, blurry, bad penis, red hair, (fur skin), gnome, bad glande, obese, brown nose, bad anatomy, colored snout, painted snout, black snout, brown snout, holding, fusion, mutation, mutilation, body hair, size difference, ((solo, solo focus)), simple background, chibi, (flat color), animal,
Notes:
-
Sometimes put in "from side"
-
The character has only "unibrow", but it almost always appears separated
-
Sometimes a "tail" is generated
-
there is no need to add a default for "size/age"
-
It's better to indicate a scenario
-
The most noticeable thing I have seen is that when a gnome size is specified, it generates a miniperson gnome character.
-
I recommend you to add a lora detail, I think it can help you. I'm not yet aware of the best prompt for this model.
描述:
I use this expression for both to interact:
masterpiece, best quality, amazing quality, colorful, vibrant, very aesthetic, detailed background, 2boys, multiple boys, age difference,<lora:Jiromaru-000008:0.95>, tbatb-jteen, 1boy, "weight", "clothes", "characteristic", "reactive action", "place", boar boy, young, "active action", "pose", "expression", "weather", etc..., BREAK (tbatb-jiromaru), 1boy, "weight", "clothes", "active action", "expression", "stature", "reactive action", "pose", etc..., <another Lora>, active Lora, <another Lora>, active Lora,
masterpiece, best quality, amazing quality, colorful, vibrant, very aesthetic, detailed background, (((2boys))), (multiple boys), (size difference:0.3-0.5), "weather", "place", "angle", together, <another Lora>, active Lora, <lora:Jiromaru-000008:0.95>, (tbatb-jiromaru), 1boy, "weight", "stature", "clothes", "pose", "individual action", "characteristic", BREAK tbatb-jteen, 1boy, "stature", boar boy, "weight", "clothes", ("individual action"), characteristic,
pose = standing/sitting/kneeling/lying, etc
characteristic = closed eyes/crossed arms/long legs/ ->body parts
It might not be the best, but it works well for me
I have read that when using a "lora xl" other than illustrous it is necessary to put a “,” before the activator and with the lora, I don't know if this is true.
训练词语: tbatb-jiromaru, 1boy, chubby,red scarf, yellow jacket, red sash, white baggy shorts, brown shoes,red scarf, orange vest, red sash, white baggy shorts, brown shoes,tbatb-jteen, 1boy, fat,red scarf, white jacket, brown sash, beige baggy shorts, brown shoes,masterpiece, best quality, amazing quality, colorful, vibrant, very aesthetic, detailed background
名称: 1198852_training_data.zip
大小 (KB): 11648
类型: Training Data
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
名称: Jiromaru-000008.safetensors
大小 (KB): 139593
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success