Stable Diffusion is a powerful tool for generating images, but to unlock its full potential, you need to have the right models or checkpoints installed. In this blog, we will guide you through the process of downloading and installing models in Stable Diffusion.
In the context of Stable Diffusion, an essential aspect lies in the utilization of models, commonly referred to as checkpoints. These checkpoints hold immense significance as they serve as pre-trained neural networks, empowering the generation of visually captivating images. By employing a diverse array of models, one gains the ability to delve into a multitude of captivating themes, artistic styles, thereby expanding the creative horizons and possibilities within the realm of image generation.
To begin the process, you need to visit the Civit AI website. It's a fantastic resource that offers a wide range of models for Stable Diffusion. Accessing their website will open up a world of possibilities for your image generation endeavors. Once you're on the Civit AI website, you can browse through the available models.
Take your time to explore different categories, such as architecture, nature, people, or even anime. This will help you find the perfect model for your specific needs and interests.
When you find a model that you want to use, simply click on the download button provided. Keep in mind that models can be quite large, typically ranging from two to seven gigabytes in size. Make sure you have sufficient disk space before initiating the download.
For this tutorial we will download the reV Animated model:
After the download is complete, you will have a model file in your downloads folder. To install the model, you need to move this file to the designated models folder in your Stable Diffusion installation directory. This process ensures that Stable Diffusion recognizes and utilizes the installed model during image generation.
Make sure you place the downloaded stable diffusion model/checkpoint in the following folder "stable-diffusion-webui\models\Stable-diffusion":
In addition to downloading and installing models, it's crucial to understand certain details that can greatly impact your stable diffusion experience. Two key aspects to consider are:
What are Pruned models? Pruned models are optimized versions that remove unnecessary data, resulting in smaller file sizes.
What are variable autoencoders (VAEs)? VAEs are responsible for generating the final image from the model's internal latent representation. You can further enhance your stable diffusion journey by adding custom VAEs. Obtaining VAEs involves finding suitable ones either through dedicated websites or platforms like Civit AI. For example, the reV Animated model we downloaded in the beginning of this tutorial recommends using one of three VAEs for better and higher quality image outputs.
Once obtained, installing VAEs and making UI modifications allow you to select and utilize them within Stable Diffusion. Different VAEs can produce varied visual results, leading to unique and diverse images. More information on how to install VAEs can be found in the tutorial listed below.
With the model successfully installed, you can now utilize it for rendering images in Stable Diffusion. The process involves selecting the downloaded model within the Stable Diffusion interface.
Model/Checkpoint not visible? Try to refresh the checkpoints by clicking the blue refresh icon next to the available checkpoints.
Congratulations! You've learned how to download and install models in Stable Diffusion. By expanding your collection of models, you can unleash your creativity and explore various artistic possibilities. Remember to experiment with different models and styles to create unique and captivating images. Happy modeling!