Stability AI has launched a new AI image generation model called Stable Diffusion XL Turbo that can create images as you type text prompts. The software is groundbreaking for its speed and interactivity.
As users input text, Stable Diffusion XL Turbo starts generating corresponding images immediately.
The images transform fluidly, adapting to each word added or removed from the text prompt.
The model utilizes novel techniques like Adversarial Diffusion Distillation and adversarial loss. These methods enhance the quality and realism of the AI-generated images compared to previous models.
The streamlined single-step image generation is also much faster than predecessor models that took 20-50 steps.
Early blind testing shows the new system can outperform other models on image quality metrics.
Stable Diffusion XL Turbo is currently available in beta testing under a free non-commercial license. The software works across devices and browsers, accessible on Stability AI’s Clipdrop platform.