New features of Stable Diffusion 3

Stable Diffusion 3 has opened applications for early access to Stable Diffusion 3 preview, in which the issue with text generation on images is expected to be finally resolved. Additionally, there should be improvements in working with multiple characters or objects simultaneously. It is reported that in the new version, arguably the most popular AI model for generating images from textual descriptions, will have access to anywhere from 800 million to 8 billion parameters, naturally expanding customization capabilities tailored to user tasks.

In the update, security was also addressed, with developers stating that they have made maximum efforts to prevent the misuse of the model. Collaborating with researchers, experts, and the community, developers aim for further innovation implementation while adhering to specific principles as the public release of Stable Diffusion 3 approaches.

As of now, Stable Diffusion 3 is not available to the general public in any form. However, there is a set of images generated by the new model, showcasing its excellent handling of text and complex images with multiple characters.