Black Forest Labs has released FLUX 3 Image, the image-generation and editing arm of its multimodal FLUX 3 family — and its headline trick is surgical precision: edit one region of a picture while, the company claims, every other pixel stays exactly as it was.
Launched October 1, 2026, FLUX 3 Image builds on the FLUX 3 family announced in July 2026. The model's signature feature is multi-step editing with what BFL describes as pixel-level fidelity outside the edited region — a direct answer to the biggest frustration with generative image editors, which tend to resample the whole frame and subtly alter faces, text, and backgrounds along the way.
Layout control is the second headline. Creators can compose scenes with bounding boxes — JSON element boxes on a normalized 0–1000 grid — so objects land exactly where they want them instead of wherever the model feels like putting them. Combined with text-to-image and image-to-image generation, text rendering, photorealism, and support for up to 10 reference images in a single prompt, it is a package aimed squarely at professional production workflows.
Resolution is the third pillar: native output up to 4K, with the vendor showing 2K and 4K samples. That is a meaningful jump for a field where 1024-pixel generations have long been the norm.
On the commercial side, API access is 50% off through October 8, 2026, and a free demo is available for anyone who wants to try it. Companies can already license commercial weights, and an open-weight release is expected in the coming weeks — a pattern BFL has followed with earlier FLUX models.
The timing is no coincidence. Ideogram announced its editing-focused v4.5 shortly before this launch, also expected as open weights soon. Two strong open-weight editors arriving within days of each other signals an image-editing arms race — good news for creators who want local, licensable control over pixels.
Our take: the 0–1000 bounding-box grid and the pixel-fidelity claim are the parts worth watching. If they hold up in real-world use, FLUX 3 Image could become the default tool for precise, iterative image work — and native 4K output removes one of the last excuses for not shipping AI-generated visuals in production.
Comments
Post a Comment