Stability AI has launched an artificial intelligence image generator that produces AI images in real-time as the user is typing a prompt.
According to a blog post, SDXL Turbo is “state-of-the-art” technology that reduces the step count from 50 to just one.
Quick test with SDXL Turbo, another amazing, super-fast diffusion model. It works right out of the box with @diffuserslib
unofficial demo txt2img and img2img: https://t.co/xdYgZpkVN4 pic.twitter.com/abMET8RuvN— Radamés Ajna (@radamar) November 29, 2023
Users input the image they want to see and while typing the image changes with each word written. Radamés Ajna posted an example video (seen above) on the platform formerly known as Twitter (now X).
The model can also use an image as a prompt or even connect to a computer’s webcam and instantly begin AI-generating elements onto the webcam image.
An unofficial SDXL Turbo demo has been posted online for users to try out.
Typically with AI image generators, it takes at least a few seconds to generate an image, sometimes a lot longer depending on the model and the server.
Stability AI — the makers of Stable Diffusion — say that SDXL Turbo is based on a “novel distillation technique called Adversarial Diffusion Distillation (ADD), which enables the model to synthesize image outputs in a single step and generate real-time text-to-image outputs while maintaining high sampling fidelity.”
“Turbo” is not available for commercial use yet but Stability AI has published a research paper which contains all the technical details.
“By incorporating ADD, SDXL Turbo gains many advantages shared with GANs (Generative Adversarial Networks), such as single-step image outputs, while avoiding artifacts or blurriness often observed in other distillation methods,” says Stability AI.
Active Company
Stability AI keeps on pushing out generative AI models: last week, it announced an AI video generator which was immediately being used to animate classic memes that were once still photos.
The company has released a litany of AI image generator models that are often adapted by other AI companies and researchers to make specific products. It was only in July when Stability AI released Stable Diffusion XL which at the time was the company’s most powerful AI image generator.