settings and stuff below
collection of (not my) NSFW LoRas!
let me know of your LoRas and i may add them
Discord: killedmyself.
V13: merged v12 with a few other different mixes. Check the example images to recognize this model's art style
Settings i Use (feel free to experiment):
-
sampler: DPM++ 2M Karras
-
steps: 30 (another 30 if using hires, or inpainting)
-
size: 640x832 (then upscale by 4, or use a hires fix)
-
cfg scale: 7-12 (just depends on what you want out of the prompt. experiment.)
-
clip skip: 2
Stuff to Note:
-
It really likes wood, so i advise adding “wood” to your negatives if you find that it makes everything out of wood.
-
Hands are usually pretty awful (obviously), so embeddings and LoRa’s or inpainting recommended to fix.
-
hires fix also recommended, or just upscale after some inpainting (what i do).
Have feedback or requests? let me know!
Looking forward to your reviews!
Please consider to support me so i can upgrade my struggling 3060, to a 40xx so i can train XL!
ko-fi: https://ko-fi.com/killedmyself
other links: https://linktr.ee/iot4
描述:
more focus on backgrounds, and lighting and shadows.
训练词语:
名称: uselessANIMERGE_v12.safetensors
大小 (KB): 4308045
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
