Phoenix crown and bridal robe, spring breeze face, ten miles of red makeup will not disappoint you.LoRa ModelThere is a LORA named after the Phoenix Crown and the Phoenix Burial Cap. In the LORA model of the LiblibAI website SDXL, the number of visits and downloads is still very high in the ancient style series. Today we will use this LORA model and combine it with the InstantID plug-in to see how to implement AIArt Photo.
1. Chinese style | Hanfu | Phoenix crown and bridal robe SDXL
Model download address
LiblibAI:https://www.liblib.art/modelinfo
/1b34e7914de04a849fbc8aa1a8358b50
The author provides instructions on how to use the LORA model
(1) Trigger word: QIEMANCN, weight value: 0.6-0.8
(2) Related prompt words: Chinese style, Hanfu;
(3) Recommended large model: Basic model SDXL 1.0
(4) CLIP skip layer: 1
(5) Output size: 1024*1280
Let’s first take a look at the output effect of this LORA model.
Prompt：High quality,masterpiece,rich details,realistic photography,8k,high-definition image quality,
QIEMANCN,a photo of woman with a red and gold dress,hanfu,chinese style,grand spectacle,wide viewing angle,high definition,big scene,snowscape,plum blossom,looking at the camera,
Prompt word: High quality, masterpiece, rich details, realistic photography, 8k, HD quality,
QIEMANCN, photo of a woman, red and gold clothes, Hanfu, Chinese style, spectacular, wide viewing angle, high definition, large scene, snow scene, plum blossoms, looking at the camera,
Reverse prompt words:
easynegative,ng_deepnegative_v1_75t,(worst quality:2),(low quality:2),(normal quality:2),lowres,bad anatomy,bad hands,normal quality,((monochrome)),((grayscale)),(( watermark)),
Related parameter settings:
- Large Model: LEOSAM HelloWorld New World | SDXL Realistic Large Model
- Sampler: DPM++2M Karras
- Sampling iteration number: 30
- Image width and height: 1024*1024
- Prompt word guidance coefficient (CFG): 7
- LORA: The weight is set to 0.8
- The effect of generating pictures is as follows.
About the LORA model testing experience
(1) The human face is prone to distortion. The author has also consulted fellow Taoists on the official website for a good solution.
(2) The hand problem is quite serious. Of course, this problem is difficult to handle with general models.
(3) The phoenix crown on the head seems too simple and not grand enough.
(4) The recommended LORA model is the SDXL1.0 standard model. It is better to try the LEOSAM HelloWorld realistic large model.
2. How to create a plug-in face-changing function using InstantID
[Step 1]: Selection of large model
We use LEOSAM HelloWorld|SDXL photorealistic large model
【Step 2】：Writing prompt words
Positive prompt words:
Prompt: High quality, masterpiece, rich details, realistic photography, 8k, high-definition image quality,
QIEMANCN,a photo of woman with a red and gold dress,hanfu,chinese style,grand spectacle,wide viewing angle,high definition,big scene,snowscape,plum blossom,looking at the camera, ,
Related parameter settings of Wensheng diagram
- Sampler: DPM++2M Karras
- Sampling iteration number: 30
- Image width and height: It is recommended to set it to around 1024.
- Prompt word guidance coefficient (CFG): 3
【Step 3】ControlNet plug-in InstantID model setting
Two ControlNet units are required here.
ControlNet Unit 0: Upload original image of a real person
Here we use photos of Fairy Sister to generate AI artistic photos.
Note that the image here should be around 1024 in size.
The relevant parameter settings are as follows:
- Control Type: Select "Instant_ID"
- Preprocessor: instant_id_face_embedding
- Model: ip-adpter_instanid_sdxl
- Control weight: 1
ControlNet unit 1: Upload face positioning photo
Upload any photo, its role is to control the final image of the character's face position. We can do the same as the first picture.
The relevant parameter settings are as follows:
- Control Type: Select "Instant_ID"
- Preprocessor: instant_id_face_keypoints
- Model: control_instant_id_sdxl
- Control weight: 0.45
【Step 4】Image generation
Click the [Generate] button and let’s take a look at the final generated image effect.
3. Change the character's posture
We can change the action posture of the character through ControlNet unit 1. Just replace the character image uploaded by ControlNet unit 1. If the effect is not obvious, the image of ControlNet unit 0 and ControlNet unit 1 remain the same.
The generated image effect is as follows.
Okay, that’s all for today’s sharing. I hope that what I shared today will be helpful to you.
The model is placed in the network disk, and those who are interested can take it!
Network disk link: https://pan.quark.cn/s/b3df771404e2