王国华

王国华

采用区块链技术建立的个人博客

EasyPhoto | Your intelligent AI photo generator

🦜 EasyPhoto is a Webui UI plugin used to generate AI portraits. The code can be used to train a digital avatar related to you. It is recommended to use 5 to 20 portrait images for training, preferably half-body photos without wearing glasses (a small amount is acceptable). After training is completed, we can generate images in the inference section. We support using preset template images or uploading your own images for inference.

Our UI interface is as follows:
Training section:

train_ui

Inference section:

infer_ui

Quick Start:

  1. Cloud Usage: AliyunDSW/AutoDL/Docker
    a. Through Aliyun DSW
    DSW provides free GPU time, which users can apply for once and is valid for 3 months after application.

Aliyun provides free GPU time in Freetier, obtain and use it in Aliyun PAI-DSW, and EasyPhoto can be launched within 3 minutes: https://gallery.pai-ml.com/#/preview/deepLearning/cv/stable_diffusion_easyphoto

DSW Notebook

b. Through AutoDL
If you are using AutoDL, you can quickly launch the Stable Diffusion webui using the provided image.

You can fill in the following information in the community image to select the desired image.

aigc-apps/sd-webui-EasyPhoto/sd-webui-EasyPhoto
c. Through Docker
If you are using Docker, make sure that the machine has the correct installation of GPU drivers and CUDA environment, and then execute the following commands:

Pull the image#

docker pull mybigpai-registry.cn-beijing.cr.aliyuncs.com/aigc/sd-webui-easyphoto:0.0.3

Enter the image#

docker run -it -p 7860:7860 --network host --gpus all mybigpai-registry.cn-beijing.cr.aliyuncs.com/aigc/sd-webui-easyphoto:0.0.3

Launch the webui#

python3 launch.py --port 7860

How to use:

  1. Model Training
    The EasyPhoto training interface is as follows:

The left side is the training image. Simply click on "Upload Photo" to upload images and click on "Clear Photo" to delete uploaded images;
The right side is the training parameters, which cannot be adjusted for the first training.
After clicking on "Upload Photo", we can start uploading images. It is best to upload 5 to 20 images, including different angles and lighting conditions. It is best to have some images without glasses. If all the images include glasses, the generated results can easily generate glasses.

train_1

Then click on "Start Training" below. At this time, we need to fill in the user ID above, such as the username, to start the training.

train_2

After the model starts training, the webui will automatically refresh the training log. If it does not refresh, click the "Refresh Log" button.

train_3

If you want to set parameters, the interpretation of each parameter is as follows:

截屏 2023-11-11 11.29.53

  1. Character Generation
    a. Single Person Template
    Step 1: Click the refresh button to query the model corresponding to the user ID after training.
    Step 2: Select the user ID.
    Step 3: Select the template to be generated.
    Step 4: Click the "Generate" button to generate the result.

single_people

b. Multi-Person Template
Step 1: Go to the settings page of EasyPhoto and set num_of_Faceid to be greater than 1.
Step 2: Apply the settings.
Step 3: Restart the UI interface of EasyPhoto.
Step 4: Go back to EasyPhoto and upload multi-person templates.
Step 5: Select the user IDs of two people.
Step 6: Click the "Generate" button to perform image generation.

multi_people_1

multi_people_2

Loading...
Ownership of this post data is guaranteed by blockchain and smart contracts to the creator alone.