Stablediffusion Introduction to the Usage of Several Core Features of SD

To learn to use SD, you need to understand the following core features, because during the creation process you will often need to debug and switch between different features to achieve the desired image output. Therefore, introductions to these features are essential.



Large models:
Large models are deep learning models generated by using materials and the SD base model, and they can be directly applied to generate images. Large models are the core material for creation and determine the direction and style of the final work. The extensions of these large models are generally CKPT or SAFETENSORS. It is recommended to use a GPU with at least 8GB of VRAM for operation.

VAE:
VAE is a supplement to large models, similar to a filter. It can stabilize the color range of the image and improve the aesthetics of the work. The extensions of VAE are generally CKPT or SAFETENSORS. It is recommended to use a GPU with no less than 8GB of VRAM for operation.

LoRA:
LoRA is a model plugin. It requires deep learning based on a certain large model to generate a small model. It needs to be used together with large models and can influence styles within a small to medium range or add elements missing from large models. If it is generated based on the SD base model, it will have better versatility and applicability when switched between different large models. The extensions of LoRA are generally CKPT or SAFETENSORS.

ControlNet:
ControlNet is a powerful plugin that gives SD the ability to analyze information such as lines and depth of field in images and map that information back onto the image being processed. This is very useful for creating truly natural and realistic images.

Stable Diffusion Web-UI (SD-WEBUI):
SD-WEBUI is open-source software made using Stability AI algorithms, and it allows you to operate SD through a browser. This open-source software not only has a complete set of plugins and is easy to use, but can also receive updates and support at any time. The SD-WEBUI runtime environment is based on Python, so some programming knowledge is required for operation.