Z Image

Z Image is offered as a browser-based image generator and editor on Fooocus.one. The site is an independent third-party platform that aggregates several image and video models, so it is not the official home of the Z-Image model itself. The tool page focuses squarely on Z-Image and its editing variant, Z-Image-Edit, packaging them into a simple prompt-to-image workflow with no installation required.

The underlying model is notable for two things. First, an emphasis on photographic realism: fine control over lighting, texture and composition so results read as photographs rather than obviously synthetic renders. Second, bilingual capability, with accurate rendering of both Chinese and English text, including small font sizes, which makes it unusually practical for posters, packaging mockups and typographic design.

Technically, Z-Image uses a Scalable Single-Stream DiT (S3-DiT) architecture that combines text, visual semantic tokens and image VAE tokens into a single sequence for the Transformer backbone, which improves parameter efficiency at around six billion parameters. Generation is fast, typically completing in a few seconds even on consumer hardware, and a built-in prompt enhancer uses a reasoning chain to interpret vague or abstract requests such as classical poetry or visual puzzles.