AI News

Stable Diffusion 3.5: What’s New and How It’s Changing the Game

0
Please log in or register to do it.
Stable Diffusion 3.5: What’s New and How It’s Changing the Game

On October 22, 2024, Stability AI announced the release of Stable Diffusion 3.5. This model comes in a:

  • Stable Diffusion 3.5 Large: This one has 8 billion parameters (so, it’s pretty smart) and can churn out images up to 1 megapixel.
  • Stable Diffusion 3.5 Large Turbo: The Turbo version is a bit of a cheat code for speed. It’s a distilled version of the Large model, but it can generate high-quality images in just 4 steps. Fast and sharp.

Coming soon—Stable Diffusion 3.5 Medium, due on October 29th. This one’s designed to run smoothly on consumer hardware, making it a bit friendlier for folks not rocking the latest GPU. It’ll have 2.6 billion parameters, which balances customization options with fast performance.

https://huggingface.co/stabilityai/stable-diffusion-3.5-large/blob/main/sd3.5_large.safetensors

https://huggingface.co/stabilityai/stable-diffusion-3.5-large-turbo/blob/main/sd3.5_large_turbo.safetensors

Model Advantages

  • Customizability: Easily fine-tune models and build applications based on customized workflows to meet your specific creative needs.
  • Efficient performance: Optimized to run effortlessly on standard consumer hardware, especially the Stable Diffusion 3.5 Medium and Stable Diffusion 3.5 Large Turbo models.
  • Diverse Output: Create images that are representative of not only a specific person, but the world as a whole with a variety of skin tones and features, without the need for extensive instructions.
  • Versatile Styles: Generate a wide range of styles and aesthetics including 3D, photographs, paintings, line art, and almost any visual style imaginable.

The Stable Diffusion 3.5 Medium outperforms other medium-sized models and offers an excellent balance of prompt fidelity and image quality, making it the perfect choice for those looking for efficient, high-quality performance.

How Does the Licensing Work?

Here’s a quick breakdown of Stability AI’s Community License:

  • Non-commercial use: Free for everyone! Perfect for personal projects, academic research, and just playing around with AI art.
  • Commercial use: If your project or business pulls in less than $1M a year, you can still use the models for free. That’s a pretty generous deal.
  • Big players: If you’re making over $1M, Stability AI asks that you reach out to discuss an Enterprise License. Fair enough, right?

ComfyUI’s Instant Response

One of the coolest things about this release is that ComfyUI didn’t waste any time. They rolled out support for SD3.5 on the same day. You can get started right away using their “missing model” feature, and the good news? You don’t need to be a coding genius to make it work. ComfyUI makes it easy for anyone to make it work.

Workflows:

ComfyUI Stable Diffusion 3.5 Simple Workflow
ComfyUI Tora Text to Video Workflow

Your email address will not be published. Required fields are marked *