
This Lora is intended to emulate the style of the 1988 movie "Willow". This Lora was not trained on characters. However using "short man" will generate someone similar to Warwick Davis. I'm new to Lora training so I'm sure there are things I can improve on, however I'm happy with the result. Use verbose natural language descriptions for best results. The more descriptive the prompt, the less it seems to adhere to the loras aesthetic, but of course you get more of what things you want to see. Lower guidance results in better realism but hands get mangled and detail is lost. Steps....I dunno, sometimes it works better by increasing or decreasing. There is no need to prompt any crap beyond what you want to see. The Lora will do the rest.
Steps: 30-60
Guidance: 2-7
Lora Weight: 0.6-1
描述:
训练词语:
名称: lora.TA_trained (1).safetensors
大小 (KB): 598408
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success