
Name inspired by the 2002 film "S1m0ne" about a non-existent digital actress. An amalgamation of created and found celebrity models merged, this S1m0ne likewise does not exist.
.5 -.7 weight is recommended, depending on the model and desired effect. You'll have more freedom prompting specific clothing on the lower end, and the upper end gets racier - this LORA doesn't take much encouraging to get horny. If you want to avoid that, lower weight and/or 'nude' with emphasis in the negative is advised.
Higher end also tends to add some ornate details to environments which can be interesting, and it bakes at the top so be aware.
_________________
Trained from 65 images generated from the previously stated merged celebrities model (1.5 base), which won't be shared because you have to call out EVERY celebrity token in a row to get S1m0ne, and it's an unwieldy mess. I like it being a 'secret sauce' anyways!
Any poor results with resemblance can most often be fixed with a higher resolution img2img or inpaint, like most LORA's, assuming you have a solid prompt.
描述:
V.1, and likely the only. Haven't tried it with the keyword 'S1m0ne' although it was in the captioning, so I know it's not necessary!
训练词语:
名称: S1m0ne.safetensors
大小 (KB): 147569
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success