
V3: Better for males/furry/sci-fi/; smaller file size. Lower weights for known chars/other lora.
V2: Somewhere in between. Less gender bias, but less effect on Flux-known chars.
V1: Most details, portrait composition. Female bias. Occasional random sword.
My own Fantasy Portrait Style, using a blend of multiple styles. Pretty damn epic: more detail, more style, more personality. It'll move images towards a more painterly style and enhance details.
Trigger: alieffp_style (V1) /alieffp_v2 (V2) /alieffp (V3)
Comparisons below:
描述:
I accidentally messed up the settings for this one and it overtrained, but I had a usable epoch!
-
Better gender balance
-
Better non-human images
-
Smaller size
For known characters, cybernetics, or anthros, use lower LoRA strength to keep their likeness more!!!
训练词语: alieffp
名称: Flux_Fantasy_Portraits_V3.safetensors
大小 (KB): 18806
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success