Black Forest Labs trains FLUX 3 on images, video, audio
Andreas Blattmann and Dustin Podell of Black Forest Labs discuss the development of generative media models, including Stable Diffusion, FLUX 3, and FLUX Kontext. The creators explain their approach to training models on visual, motion, and audio signals together to build visual intelligence and improve temporal consistency. They also share insights from their transition from open-source community developers and academic researchers to founding a dedicated lab. Open article →