
An attempt to port old SD1.5 embedding to a LORA. Result is not as versatile, but much easier to use and since it's trained on Pony checkpoint, it will spawn accurate ladies with or without clothing.
Trained on Rogue Pony (which is already in the ballpark of illustration style) and is intended to use with this particular checkpoint. It may or may not work with other checkpoints. Weights between 0.5 and 1.0 work best for me, I'd start at 0.8 and tune from there.
The name is not particularly accurate, bit it has some modernistic vintage feel.
There is a trigger phrase, but it doesn't seem to matter much, I usually omit it. Also it seems to work better without "score" tags,
I trained the set on SDXL too, maybe will post it someday.
描述:
Use with Rogue Pony checkpoint.
训练词语: retfut_style
名称: retrofuturistic_pony_style_rogue.safetensors
大小 (KB): 72392
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success