
This lora is based on kohaku zeta XL training
This is my first attempt to make XL's painting style lora, because for me, painting style is a very abstract thing, especially the different brush styles of the same artist will change with time, and my naked eye can not accurately judge which kind of painting style lora belongs to, so I rarely make painting style Lora
I have been trying to find open source tools that can identify the style of painting to help me judge, but unfortunately most of them do not meet my requirements, and finally I decided to use CSD to help me with this operation. This model uses CSD to cluster the painting styles of the same artist in different periods, so that we can call the specific painting styles of this artist instead of a certain general range
Who is Artist 001
He is an excellent artist, and for some reason I think using code names instead of artist names is also convenient for me to keep track of the number of artists
Trigger words:
1.art001
2.A_style
3.B_style
4.C_style
5.D_style
after the classification of CSD 001 painting style can be roughly divided into 5 types, you can use trigger words and corresponding labels to change out of a specific painting style, such as art001+A_style
Although maybe it doesn't make that much difference
It should be noted that the first edition of this lora does not require the use of additional trigger words, the second edition has been trained by some data set screening and adding the trigger word art001, while the third edition is the product of balancing the number of different painting style training sets of the artist and adding the corresponding style words (A_style, etc.)
Recommend weight:
0.8-1
I also plan to make this artist lora on other large models in the future. If you have better painting style classification tools or better marking methods to distinguish different painting styles of the same artist, please feel free to share
描述:
训练词语:
名称: art001_styleV1.safetensors
大小 (KB): 68870
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success