Special note: Please do not add the trigger word hunsha_girl after loading Face Lora

The author himself forgot to delete the trigger word during the test and thought there was a problem. Special reminder!

Otherwise, contamination will occur, and the face shape will not match!!

This model can not only generate a girl in a wedding dress, but the realistic effect is also pretty good.

All kinds of gameplay are welcome to test! Everyone is welcome to use it!!!

QQ group number: 833273550 (model test map can be downloaded in the group for everyone to choose parameters)

Can be used with all of my filters

link

https://www.liblib.ai/userpage/aa347406fabf4c55a71e9308889f0db4/model

Friends are welcome to use it! Back to the picture!!

TikTok 63223961125 Look forward to everyone's attention!

Bottom film selection during training Big Orange v7 Train.

Thanks to the boss and thank the owner of the website B Chen Erha

's technical support!

[Model introduction] I wanted to make this model to have girlfriends who wear all kinds of wedding dresses. Currently, only girls' single-person perspective images can be generated, and they will slowly be improved in the future.

[Generation suggestions] 1. First, it is recommended to use it with your own face model. The number of steps is 28-50. The sampler suggests dpm++2m sde karras (the results are the most stable after testing; feedback from bosses is welcome). The cfg value is recommended to stay at 7, and the basic resolution : 512*768, amplification algorithm r-esrgan4x+ , Redraw amplitude: 0.2-0.4, Magnification: 2

[Creative Overview] This model aims to allow more girls to wear all kinds of wedding dresses to show off their most beautiful self, but it's not limited to that.

Share a set of tags I used for testing for everyone to use:

Positive Reminder: 1girl, dress, solo,

flower, white dress, black hair, bouquet, jewelry, wedding dress, looking at viewer, blurry background, white flower, realistic, holding, hair bun, black eyes, negative Reminder words:

(worst quality: 2), (low quality: 2), (normal quality: 2), lowres, watermark,