
This LoRA was trained on about 50 screenshots from the television series Xena: Warrior Princess in 480p. Training was heavily focused on trying to eliminate the butt chin, with some success though it may still pop up from time to time.
Recommend strength for the 640 version is 1.0-1.2, though the latter might give a slightly better likeness.
Recommended strength for the 128 version is 1.5.
This model was trained by finetuning a Flux1D checkpoint with kohya_ss and then using the lora extraction script to extract the difference between it and the original model as a 640dim model. It is the most exact version of Xena possible without the checkpoint. Luckily the 128dim 1.1gb model works very well as well.
Like my work? Please consider leaving a tip to help fund future LoRAs or commission a custom LoRA for yourself. Higher quality versions of my LoRAs are also available for a monthly donation: https://ko-fi.com/artfulgenie.
描述:
This is the 128 version of the LoRa, for those who need a smaller version. Recommended strength of this version is 1.5.
训练词语: xena,xena-eyes,xena-style
名称: flux_xena_128_newest.safetensors
大小 (KB): 1196654
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success