
The Pony version of the lora I made. This one is not as good as the 1.5-version, but still OK. It won't generate breeches for some reason. It just gives long pants or just stockings... I made 2 versions.
! WARNING: USE THE CORRECT VERSION !
-
V0: a version trained on the default Pony model. Gives good results with the base model but gives problems when used with realistic Pony checkpoints. It will keep replicating the wigs, but clothing will be simpler and the makeup will be ignored...
-
V1: a version for realistic checkpoints, this is a mix of the first model with a failed version trained on the model Stallion Dreams. This mix generates better clothing and makeup with realistic pony models than the standard version. Results vary depending on the checkpoint.
This lora tries to capture the look of French noblemen of the 18th century. My lora focuses mainly on the wigs and makeup, but it still does a decent job with the clothing (except for breeches). Prompting is still required to get all aspects of the look right. Use "historical wig, grey/white hair, 18th century clothes, coat, waistcoat, breeches, white stockings, red lips, white face makeup, pink cheeks, rouge" in your positive prompt and "lipstick,facial hair,beard,stubble" in the negative prompt.
I used a small dataset of only 20 realistic images (AI edited paintings and photos). All images of the dataset depict handsome men with wigs and subtle period makeup. Don't expect full historical accuracy, (or breeches).
My SD1.5 version: https://civitai.com/models/1071861/18th-century-men-rococo-clothing-makeup-wigs
描述:
Use with standard Pony model
训练词语: W16hsL1ps,1boy
名称: 18th_Cent_man_PXL_V0.safetensors
大小 (KB): 111755
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success