NVIDIA has released the ultra-compact — Perfusion AI
<p>NVIDIA researchers recently introduced a text-to-image model called Perfusion, which stands out for its compact size, requiring only 100KB of space and a training time of just 4 minutes. Despite its small size, the generated images are still impressive.</p>
<p><iframe frameborder="0" height="480" scrolling="no" src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FWhFoEaB7CvY%3Ffeature%3Doembed&display_name=YouTube&url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DWhFoEaB7CvY&image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FWhFoEaB7CvY%2Fhqdefault.jpg&key=a19fcc184b9711e1b4764040d3dc5c07&type=text%2Fhtml&schema=youtube" title="Stable Diffusion with Better Control! Perfusion Model Explained (by NVIDIA)" width="854"></iframe></p>
<p>NVIDIA collaborated with researchers from Tel Aviv University in Israel to develop the Perfusion model. Officially stated, the generated image quality of the Perfusion model is comparable to Stability AI’s Stable Diffusion v1.5 version. However, Perfusion stands out in efficiency due to its lightweight design, placing it at the forefront of the industry in terms of efficiency.</p>
<p><a href="https://generativeai.pub/nvidia-has-released-the-ultra-compact-perfusion-ai-5df557b833c9"><strong>Website</strong></a></p>