V2：

The new LoRA takes another step forward in terms of character flexibility and versatility, but the accuracy of his original outfit has decreased a little.Since the training dataset is still relatively small, the results can sometimes be a bit inconsistent. I may continue adding more training images in the future to improve its stability and overall performance!

Promote:

Jan_dolski , bright blue eyes , mustache , stubble ,

Some Tips for Using the LoRA

• Face detailing: When using a face/detailer pass, try adding caucasian, pale skin to the prompt.

• Jan's original outfit: If you want to recreate Jan's original clothing, try using jacket, red trousers.

• Hands: The hand details in the training images aren't very good, so you may need to do an additional hand-detailing pass or add prompts such as detailed hands.

Original:

Please note that this LoRA was trained using only images of Jan in his original appearance. I also didn’t tag his specific physical features separately during the captioning process.

Because of this, it may not be the best choice if you want to generate more diverse variations of Yang while changing his appearance.

Promote:jang_dolski,

middle-aged male, caucasian, pale skin, bright blue eyes, facial moles,thin body,fit body,lean male,mustache,stubble,

I’ll try making another version with more detailed tagging, so it can better handle the different variations of Jan!