Stability ai introduces SDXL Turbo, which represents a remarkable advancement in text-to-image synthesis, powered by an innovative distillation method known as Adverse Diffusion Distillation (ADD). This advancement allows the model to quickly generate high-fidelity image results, reshaping the approach to real-time text-to-image conversion.
SDXL Turbo, an evolution of its predecessor SDXL 1.0, introduces ADD, a distillation technique that combines adversarial training and score distillation. This innovative approach allows the model to generate text-to-image results in real-time with unparalleled fidelity, preserving quality and dramatically reducing the number of steps required from 50 to just one. For a deep understanding of the technical complexities, the research work delves into the particularities of this innovative distillation technique.
In particular, SDXL Turbo's ADD offers several key advantages reminiscent of generative adversarial networks (GANs), such as one-step image synthesis, avoidance of common artifacts, and blurring seen in other distillation methodologies. The article clarifies this novel distillation technique and highlights its impact on real-time image generation.
Performance evaluations performed against several variants of the diffusion model (StyleGAN-T++, OpenMUSE, IF-XL, SDXL and LCM-XL) underline the supremacy of SDXL Turbo. In blind tests evaluating cue fidelity and image quality, the SDXL Turbo outshone a 4-step LCM-XL setup with a single step. It even topped a 50-step SDXL setup with just four steps. These results accentuate the remarkable performance of SDXL Turbo, outperforming state-of-the-art multi-step models with significantly reduced computational demands while preserving superior image quality.
Additionally, the inference speed achieved by SDXL Turbo is noteworthy. On an A100, the model generates a 512 × 512 image in just 207 ms (fast encoding + single step denoising + decoding, fp16), with only 67 ms attributed to a single direct UNet evaluation.
To experience the capabilities of SDXL Turbo firsthand, people can explore real-time imaging through clipdrop, the image editing platform. The beta demo shows SDXL Turbo's prowess at transforming text prompts into stunning visual results. Clipdrop can be accessed in most browsers and offers a free trial to explore the cutting-edge capabilities of SDXL Turbo.
Review the ai/news/stability-ai-sdxl-turbo” target=”_blank” rel=”noreferrer noopener”>Model, ai/news/stability-ai-sdxl-turbo”>Reference article, and Manifestation. All credit for this research goes to the researchers of this project. Also, don't forget to join. our 33k+ ML SubReddit, 41k+ Facebook community, Discord channel, and Electronic newsletterwhere we share the latest news on ai research, interesting ai projects and more.
If you like our work, you'll love our newsletter.
Niharika is a Technical Consulting Intern at Marktechpost. She is a third-year student currently pursuing her B.tech degree at the Indian Institute of technology (IIT), Kharagpur. She is a very enthusiastic person with a keen interest in machine learning, data science and artificial intelligence and an avid reader of the latest developments in these fields.
<!– ai CONTENT END 2 –>
