Original Character LoRA. Still playing with datasets and how best to get a consistent, quality character output while also getting a LoRA that plays nice with other Style/Action/Pose LoRAs. Generally happy with Version 2 at the moment!
Version 2 has somewhat less embedded style and seems to stay closer to the original character over many tests alongside clearer prompt adherence. I mainly use 0.9 strength and just recommend "small breasts" to keep it on track.
Version 1 works between 0.7 - 1. I recommend using "(turquoise eyes:0.85)" alongside the LoRA to get the desired output.
描述:
训练词语:
名称: 800371_training_data.zip
大小 (KB): 43803
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: Pala_N_-_2.safetensors
大小 (KB): 223106
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
