Robin Rombach is a German machine learning researcher known as the lead author of the latent diffusion model architecture that became the technical foundation of Stable Diffusion, and as a co-founder of Black Forest Labs, the company behind the FLUX image-generation models.
Rombach conducted his research as a PhD student in the CompVis (Computer Vision) group at Ludwig Maximilian University of Munich, under professor Bjorn Ommer. In 2022 he was lead author, alongside Andreas Blattmann, Dominik Lorenz, Patrick Esser and Ommer, of the paper "High-Resolution Image Synthesis with Latent Diffusion Models," presented at CVPR 2022. The paper's key contribution was running the diffusion denoising process in a compressed latent representation produced by an Autoencoder rather than directly on pixels, which cut the compute required to train and run high-resolution image generators by a large factor. That efficiency gain made it practical to train and publicly release an open-weights model capable of running on a single consumer GPU, work that Stability AI funded and released publicly in August 2022 as Stable Diffusion.
From Stability AI to Black Forest Labs
Rombach subsequently joined Stability AI, where he worked as a lead researcher on later Stable Diffusion releases, including the Stable Diffusion 3 model family. In early 2024 he left Stability AI, alongside fellow latent diffusion co-authors Patrick Esser and Andreas Blattmann, to found Black Forest Labs, an independent German AI lab focused on generative media models, taking its name from the Black Forest region of southern Germany.
Black Forest Labs released the FLUX family of image generation models in August 2024, positioning them as an open and commercially licensed alternative to closed Text-to-image generation systems. FLUX was quickly adopted across the open-source generative AI ecosystem and integrated into products such as Grok's image generation features. The lab has since expanded into video generation research, competing in a field that also includes Runway, Kling and Seedance.
Significance
Rombach's work is widely credited as one of the pivotal technical contributions that made high-quality diffusion-based image generation broadly accessible outside a handful of well-funded labs, in a manner comparable to the role the Transformer (architecture) architecture played in language models. His path from an academic lab, to a funded startup, to founding his own company reflects a broader pattern of researchers whose papers underpin major commercial AI products going on to build companies around that work.