Google proposes Diffusion Controller for simpler image generation control

On September 30, 2026, Google Research announced a method called Diffusion Controller, presented as a way to simplify how control is handled in AI image generation. According to the announcement, the approach unifies and simplifies spatial and semantic control in image generation, which the team describes as improving generation precision and editing flexibility. The item appeared under Google Research's Algorithms & Theory category. Nothing beyond that was disclosed in the announcement: no specific model or version number, no benchmark results, and no code, release date, or availability details were included, and no researchers or partner organizations were named. As of this report, the work remains at the point of a method being introduced, with no further progress made public.