Stable diffusion is a large model for generating images from text. It can generate pictures using different kinds of models, each with specific functions, so as to realize the customization of different details.
1. Checkpoint model
The first is the CheckPoint model, which is the main model of stable diffusion and contains a lot of scene materials, so it is very large, and other models are customized with some details based on it.
2. lora model
The second model is the lora model, which is a fine-tuned model. It is mainly used to customize characters. Compared with the main model, the lora model is lighter and more efficient in training.
3. VAE model
The third model is the VAE model, which is a beautification model. The VAE model is mainly used to beautify the color of the picture. Many master models already have this functionality built in.
4. Embedding model
The fourth model is the Embedding model, which is an embedded model. The main function of the Embedding model is to train the ability to understand text.
5. Hypernetwork model
The last is the Hypernetwork model, which is a super network model. The main function of the Hypernetwork model is to customize the style and style of the generated pictures. By using the Hypernetwork model, more detailed style adjustments and customizations can be made to the generated pictures.
