Stability ai's newest model for image generation is ai/news/introducing-stable-cascade”>Stable waterfall It promises to be faster and more powerful than its industry-leading predecessor, Stable Diffusion, which is the foundation of many other text-to-image generation ai tools.
Stable Cascade can generate photos and offer variations of the exact image you created, or attempt to increase the resolution of an existing image. Other text-to-image editing features include Inner and Outer Paint, where the model will edit only a specific part of the image, as well as Canny Edge, where users can create a new photo simply by using the edges of an existing image.
The new model is available. ai/StableCascade”>on GitHub for researchers but not for commercial use, and offers more options even as companies like Google and even Apple launch their own imaging models.
Unlike Stability's flagship Stable Diffusion models, Stable Cascade is not one big language model: it is three different models that are based on the sausage architectureThe first stage, stage C, compresses the text prompts into latents (or smaller code fragments) which are then passed to stages A and B to decode the request.
Splitting requests into smaller bits compresses the request to require less memory (and fewer hours of training on those hard-to-find GPUs) and run faster. at the same time that it works better “both in prompt alignment and in aesthetic quality.” It took about 10 seconds to create an image, compared to 22 seconds for the SDXL model currently used.
Stability ai helped popularize the stable diffusion method and has also been the subject of several lawsuits alleging that Stable Diffusion was trained on copyrighted data without the permission of rights holders; A lawsuit by Getty Images in the United Kingdom against Stability ai is scheduled to go to trial in December. It began offering commercial licenses through a subscription in December, which the company said was necessary to help fund its research.
