In order to get ahead of its competitors in the field of productive artificial intelligence, where competition has been increasing recently, Stability AI announced Stable Cascade, which requires less computing power and can follow commands better.
Stable AI is currently Stable diffusion It is the market leader in the field of visual producing artificial intelligence. Stable Diffusion is also the basis of many other artificial intelligence tools that generate images from text. the new version Stable Cascade Together with Stability, it aims to increase the skills and efficiency of artificial intelligence.
Not only will you create visuals
Stable Cascade, can create visuals and will be able to reveal variations of the image it creates. It will also be used to increase the resolution of an existing image. Among its other features, the created image is not only editing a certain part, There are also features such as creating new images from old images.
The new model is currently available to researchers on GitHub, but commercial use hasn’t started yet. On the other hand, Stable Cascade also has an insightful structure for companies such as Google and even Apple for their own visual creation and productive artificial intelligence models. Moreover, from a technical perspective, it is quite different. Not a single large model like Stable Diffusion, Stable Cascade is combined with Würstchen architecture from three different models is formed.
According to the statement from Stable AI, in this system, commands are divided into smaller parts. Thus, artificial intelligence can follow commands better and requires less processing power. While the SDXL model (current model) creates an image in 22 seconds, this time with the new model to 10 seconds It’s coming down.
Source :
https://www.theverge.com/2024/2/14/24073253/stablity-ai-image-generation-stable-cascade-diffusion-model