Stability AI reveals SDXL Turbo: A Real-Time Text-to-Image Generation Model

Stability AI released SDXL Turbo, a new text-to-image mode that utilizes a distillation technique called Adversarial Diffusion Distillation (ADD). This technique allows the model to generate real-time text-to-image outputs in a single step while maintaining high sampling fidelity. SDXL Turbo builds upon the foundation of SDXL 1.0 and incorporates advancements in diffusion model technologies. Although the research paper detailing SDXL Turbo’s distillation technique is available for those interested, it is important to note that it is not yet intended for commercial use.