SD3 - Local Install Guide! FASTEST Way to run the new Model - Stable Diffusion 3

Olivio Sarikas
12 Jun 202406:15

TLDRIn this video, you will learn how to install and run Stable Diffusion 3 on your local computer. The guide covers downloading and updating the necessary files, including signing a free license for non-commercial use. Detailed instructions are provided for setting up models in both Automatic1111 and ComfyUI, including handling potential issues and loading various workflows. The video also includes a demonstration of creating prompts and generating images using the new model, with practical tips for optimizing image quality. Stay tuned for more tips and tricks in future videos.

Takeaways

  • 😀 Stable Diffusion 3 medium is released and the video will guide you through downloading and running it on your computer.
  • 📷 The images shown are first-roll renders with Stable Diffusion 3, and the prompts are not improved yet.
  • 📝 To use Stable Diffusion 3, you need to visit Hugging Face and sign a free license for non-commercial use.
  • 🔍 For commercial use, contact Stability AI to get a commercial license.
  • 📚 Download the 'sd3 medium including clip save tensor' file, which is around 6 GB, for the text encoder.
  • 💾 Download the model into your models folder for automatic 1111 or for com UI if you prefer.
  • 🔧 In com UI, update it first to be able to use the new model, which may involve updating through the com UI manager.
  • 🛠️ If the torch Cuda model breaks after updating, fix it by running the 'update com UI and python dependencies' file.
  • 🌐 Download and try different workflows available in com UI, such as basic, multi-prompt, and upscaling workflows.
  • 📜 There's a 'sd3 demo prompts txt' file with various prompts to test the model.
  • 🎨 The video demonstrates a Tex to image workflow with a prompt example, showcasing the model's creative interpretation of text.

Q & A

  • What is the Stable Diffusion 3 medium model?

    -The Stable Diffusion 3 medium model is a version of the AI model that does not include the text encoder, whereas the 'sd3 medium including clip save tensor' file, which is around 6 GB, is recommended as it includes the text encoder for better functionality.

  • Where can I find the Stable Diffusion 3 model?

    -You can find the Stable Diffusion 3 model on Hugging Face, where you need to sign a free license for non-commercial use.

  • What is the difference between the 'sd3 medium safe tensor' and 'sd3 medium including clip save tensor'?

    -The 'sd3 medium safe tensor' does not include the text encoder, while the 'sd3 medium including clip save tensor' is a 6 GB file that does include the text encoder, making it more suitable for use.

  • What is the purpose of signing a license on Hugging Face?

    -Signing a license on Hugging Face is necessary for using the Stable Diffusion 3 model for non-commercial purposes. For commercial use, one must contact Stability AI directly.

  • How large is the 'sd3 medium including clip and T5 XXL fp8' model?

    -The 'sd3 medium including clip and T5 XXL fp8' model is approximately 11 gigabytes in size.

  • What is the recommended workflow for using the Stable Diffusion 3 model in comu I?

    -The recommended workflow for using the Stable Diffusion 3 model in comu I is provided by comy Anonymous, which includes settings like using the sgm uniform scheduler with 30 steps and a CFG value of 5.5, along with the uler sampler.

  • What should I do if comu I stops working after updating?

    -If comu I stops working after updating, you should go to the comu I windows portable folder, find the 'update comu and python dependencies' file, and run it to fix the issue.

  • How can I test the Stable Diffusion 3 model with different prompts?

    -You can test the Stable Diffusion 3 model with different prompts by using the 'sd3 demo prompts txt' file provided, which contains multiple prompts to try out.

  • What is the significance of the text encoder in the Stable Diffusion 3 model?

    -The text encoder is significant as it allows the model to better understand and process text prompts, leading to more accurate and relevant image generation.

  • How can I get better quality images from the Stable Diffusion 3 model?

    -To get better quality images, you can test different settings in the comu I workflow, such as changing the scheduler, steps, and CFG value, and use the advice provided in the video for testing and optimization.

  • What is the role of the 'comu I manager' in setting up the Stable Diffusion 3 model?

    -The 'comu I manager' is an external extension that helps in updating comu I and all custom notes, ensuring that the software is up to date for using the Stable Diffusion 3 model.

Outlines

00:00

😀 Introduction to Stable Diffusion 3 Medium Download and Setup

The speaker introduces the Stable Diffusion 3 Medium model and guides the audience through the process of downloading and setting it up on their computers. They begin by directing users to Hugging Face to sign a free license for non-commercial use, with an option for commercial use upon contacting Stability AI. The speaker emphasizes the importance of choosing the correct model file, recommending the 'sd3 medium including clip save tensor' file over the 'sd3 medium safe tensor' due to the inclusion of the text encoder. They also mention the availability of different workflow files and a demo prompts text file for testing the model. The speaker then explains the steps to update Comfy UI (comu), troubleshoot potential issues, and load the workflows for image generation.

05:03

😃 Testing Stable Diffusion 3 Medium with a Creative Prompt

In this paragraph, the speaker shares their experience testing the Stable Diffusion 3 Medium model using Comfy UI. They demonstrate how to load the model, set up the workflow, and input a creative prompt for generating an image of a cat holding a sign with the text 'I love you.' The speaker highlights the model's ability to interpret and creatively respond to the text prompt, resulting in an image that includes a heart symbolizing 'love.' The audience is encouraged to like and subscribe for more content, and the video concludes with a sign-off and background music.

Mindmap

Keywords

💡Stable Diffusion 3

Stable Diffusion 3 is a new model in the field of AI image generation. It is an evolution of the previous Stable Diffusion models, designed to create high-quality images from textual prompts. In the video, the host discusses how to download and run this model on a computer, showcasing its capabilities with rendered images and emphasizing that these are first-roll results without improved prompts.

💡Hugging Face

Hugging Face is a platform that hosts a wide range of AI models, including Stable Diffusion 3. In the script, the host instructs viewers to visit Hugging Face to sign a license agreement for the model, which is free for non-commercial use. This step is crucial for legal access and use of the AI model.

💡License

A license in this context refers to a legal agreement that allows users to use the Stable Diffusion 3 model. The video mentions that viewers need to sign a free license for non-commercial purposes. For commercial use, one must contact Stability AI directly, indicating the importance of adhering to usage terms.

💡Model Versions

The script refers to different versions of the Stable Diffusion 3 model available for download, such as 'sd3 medium safe tensor' and 'sd3 medium including clip save tensor.' Each version has specific features and file sizes, with the latter including a text encoder for more advanced image generation capabilities.

💡Comfy UI (com UI)

Comfy UI, often abbreviated as com UI, is a user interface for running and managing AI models like Stable Diffusion. The video demonstrates how to use com UI to update and run the Stable Diffusion 3 model, highlighting its importance in the setup process.

💡Workflows

Workflows in the context of the video are pre-configured sets of operations within com UI that dictate how the AI model processes inputs to generate images. The host mentions different workflows such as 'basic,' 'multi-prompt,' and 'upscaling' that can be downloaded and used within com UI to test the model.

💡Update

Updating is a necessary step to ensure that com UI and its components are functioning correctly with the new Stable Diffusion 3 model. The script describes updating com UI through the manager extension and manually updating Python dependencies if issues arise.

💡Checkpoint

In the context of AI models, a checkpoint refers to a specific version or state of the model saved for use during the image generation process. The video instructs viewers to load a checkpoint, such as 'sd3 medium including clip save tensor,' into com UI to begin image generation.

💡Prompts

Prompts are textual descriptions provided to the AI model to guide the creation of an image. The script mentions 'sd3 demo prompts txt,' a file containing various prompts for testing the model's capabilities, and the host's own example prompt, 'cat holding a sign with the text I love you.'

💡Image Generation

Image generation is the process by which the AI model creates images based on provided prompts. The video demonstrates this process using Stable Diffusion 3, showing the results and discussing the creative decisions made by the model, such as adding a heart to the 'I love you' sign in the generated image.

💡Settings

Settings in this context refer to the adjustable parameters within com UI that influence how the AI model generates images. The video mentions specific settings like 'sgm uniform scheduler,' '30 steps,' 'CFG value of 5.5,' and the 'uler sampler,' which are used to fine-tune the image generation process.

Highlights

Introduction to the Stable Diffusion 3 medium model and its capabilities.

Instructions on downloading Stable Diffusion 3 from Hugging Face and signing the license for non-commercial use.

Difference between the 'sd3 medium safe tensor' and 'sd3 medium including clip save tensor' models.

Suggestion to use the 'sd3 medium including clip save tensor' file for better results.

Mention of an alternative larger model 'sd3 medium including clip and T5 XXL fp8'.

Explanation of downloading the model into the 'models' folder for automatic 1111 or for com UI.

Accessing and downloading different workflows in the 'comu I example workflows' folder.

Availability of 'sd3 demo prompts txt' for testing the model.

Guidance on updating com UI to use the new Stable Diffusion 3 model.

Troubleshooting steps if com UI fails to start after updating.

Importance of updating com UI's Python dependencies to fix issues.

Loading workflows in com UI after all updates and fixes are complete.

Introduction of comy Anonymous and his contribution to different workflows for the model.

How to load and customize the workflows for the 'sd3 medium including clip save tensor' model.

Settings recommended by comy Anonymous for the workflow, including scheduler and sampler.

Example of a creative output from the model using the prompt 'cat holding a sign with the text I love you'.

Encouragement to like, subscribe, and watch more videos for further insights.