
It can be fully used for a custom avatar side hustle
Face swapping with one picture, and various styles can be customized
Pulid is used to fix the face
So as to transfer the face to a specific fixed style
[Image]
Usage 1: Upload a style image and a frontal face photo, click generate
The workflow will reverse-engineer the style image's prompt, so you don't need to input prompts
[Image]
Usage 2: Turn off the style imitation
Upload a frontal face photo, write a prompt, and click generate
This only uses Pulid's facial imitation
The output style mainly depends on the model's influence, so it's best to add a suitable LORA model and prompts
[Image]
[Image]
Step 1: After the output is completed, you can enable steps 2 and 3 for high-resolution upscaling
This will make the person look more realistic, especially for real photos
[Image]
These two parameters adjust the degree of style image and portrait imitation
Adjust them to achieve the best results
[Image]
Crop face is for cropping the face and inputting it into Pulid as a reference
If detection issues occur, try switching to another model
[Image]
You can also join Ashuo's Knowledge Planet ID: 17259412
Direct link: https://t.zsxq.com/TIfva
For details, consult via WeChat: yumengahsuo
The community provides:
1. Exclusive advanced workflows for ComfyUI
2. Solutions to various challenging issues
3. Project requirement production
4. The latest overseas AI information
5. Community Q&A
Thank you for your support 🙇
It can be fully used for a custom avatar side hustle
Face swapping with one picture, and various styles can be customized
Pulid is used to fix the face
So as to transfer the face to a specific fixed style
[Image]
Usage 1: Upload a style image and a frontal face photo, click generate
The workflow will reverse-engineer the style image's prompt, so you don't need to input prompts
[Image]
Usage 2: Turn off the style imitation
Upload a frontal face photo, write a prompt, and click generate
This only uses Pulid's facial imitation
The output style mainly depends on the model's influence, so it's best to add a suitable LORA model and prompts
[Image]
[Image]
Step 1: After the output is completed, you can enable steps 2 and 3 for high-resolution upscaling
This will make the person look more realistic, especially for real photos
[Image]
These two parameters adjust the degree of style image and portrait imitation
Adjust them to achieve the best results
[Image]
Crop face is for cropping the face and inputting it into Pulid as a reference
If detection issues occur, try switching to another model
[Image]
You can also join Ashuo's Knowledge Planet ID: 17259412
Direct link: https://t.zsxq.com/TIfva
For details, consult via WeChat: yumengahsuo
The community provides:
1. Exclusive advanced workflows for ComfyUI
2. Solutions to various challenging issues
3. Project requirement production
4. The latest overseas AI information
5. Community Q&A
Thank you for your support 🙇