Stability AI has officially released the new version of Stable Diffusion 3.5 Large, with three new ControlNet functions added, namely Blur, Canny, and Depth. This update provides users with more powerful image generation capabilities, making the generated images richer and more diverse.

According to the introduction by Stability AI, the release of this model is based on its community license agreement, allowing users to use it for free on the premise of non-commercialization. Both individuals and institutions can use this model in fields such as scientific research without having to pay any fees. In addition, for small businesses and creators with an annual income of less than $1 million, Stability AI also allows them to use this model for commercial purposes free of charge. Such lenient usage terms have attracted a large number of users to try and apply this technology.

It is worth mentioning that when using the media works generated by Stable Diffusion 3.5, users can retain their ownership and do not need to worry about complicated authorization issues. This measure has greatly lowered the usage threshold, enabling more creators to utilize this advanced technology for creation.

Blur: Achieves extremely high-fidelity upscaling, including resolutions of 8K and 16K. It is highly suitable for tiling low-resolution images into large, detailed visual effects.

Canny: The generated image is constructed by utilizing Canny edge mapping. This control type is especially useful for illustrations but can be adapted to all styles.

DepthFM: The depth map generated by DepthFM is used to guide image generation. It is highly suitable for architectural rendering and 3D asset texturing, as well as other use cases that require precise control of image synthesis.

In an ELO comparison study of the same control type involving approximately 150 participants, Stable Diffusion 3.5 Large ControlNets ranked first in user preference among similar models.

When releasing this model, Stability AI also expressed its emphasis on security. The company has taken a series of measures to ensure that the development and use of Stable Diffusion 3.5 will not be misused by lawbreakers. This shows that the company is actively fulfilling its social responsibilities while promoting technological progress.

In addition, Stability AI has also stated that more ControlNet models will be launched in the future, including the Stable Diffusion 3.5 Medium (2B) variant and new control types. Users can look forward to more functions and application scenarios. To obtain the latest information, users can follow Stability AI through social media platforms or join its Discord community to communicate with other users.

The release of Stable Diffusion 3.5 Large has not only enhanced users' creative capabilities but also injected new vitality into the development of artificial intelligence image generation technology.

Official Introduction: https://stability.ai/news/sd3-5-large-controlnets