Black Forest Labs Launches with Advanced AI Text-to-Image Models

Black Forest Labs , founded by the team behind Stable Diffusion, has launched its advanced text-to-image AI model, FLUX.1, which reportedly outperforms Midjourney. FLUX.1 sets new standards in image generation. The company also plans to expand into video generation, aiming to compete with OpenAI's Sora, Runway's Gen-3 Alpha, and Kuaishou's Kling.

About the New FLUX.1 Model

• FLUX.1 models reportedly outperform Midjourney and DALL-E in image quality and text adherence, with a focus on accurate human anatomy.
• FLUX.1 shows competence in generating human hands, a common challenge in earlier models like Stable Diffusion.
• The models use a "hybrid architecture" combining transformer and diffusion techniques, scaled up to 12 billion parameters.
• FLUX.1 comes in three versions: a high-end commercial "pro" version, a mid-range "dev" version with open weights, and a faster "Schnell" version.

Source: LinkedIn, Ars Technica ,X

B.O