Stable Diffusion 3.5 update

Stability AI recently introduced three new ControlNet models for Stable Diffusion 3.5 Large: Blur, Canny and Depth. These models, available for commercial and non-commercial use under the Stability AI Community License, can be downloaded from Hugging Faceand their code is from GitHub. In addition, they are supported in the Comfy UI.

Stable Diffusion 3.5 Model Description:

• Blur: Provides high-quality image scaling to 8K and 16K resolutions, ideal for converting low-quality images into detailed visuals.

• Canny: Uses Canny contour maps to structure the generated images, which is particularly useful for illustrations, but applicable to other styles as well.

• Depth: Applies depth maps created with DepthFM to control image generation, which is relevant for architectural visualizations and 3D object texturing.

In a study with about 150 participants, ControlNet models for Stable Diffusion 3.5 Large ranked first in user preference among similar models.