Open Source Flux 3 Image Offers Precise AI Image Editing

Black Forest Labs has introduced FLUX 3 Image, an AI image-generation model that empowers users with enhanced control over composition, editing, and reference images. This model, built on Black Forest Labs' FLUX 3 foundation model, offers image generation capabilities up to 4K resolution.

One of the key features of FLUX 3 Image is the ability to define object placements within an image using bounding boxes. Users can specify the exact location of individuals, buildings, or other objects before the image generation process. Black Forest Labs claims that the model maintains these requested positions while producing a coherent final image, granting users greater control over the image composition beyond mere text prompts.

FLUX 3 Image also allows users to incorporate up to 10 reference images in a single generation. The model incorporates these images to influence the final result, while users can designate where objects from each reference should appear in the generated image. The company has demonstrated the model's capability to generate images containing Japanese text, including vertical writing.

A notable aspect of FLUX 3 Image is its pixel-level image editing functionality. Users can make changes to selected parts of an existing image, ensuring that the original image remains unaltered outside the edited region. This feature minimizes unwanted changes to areas the user does not intend to modify, making FLUX 3 Image a potentially useful tool for workflows that require precise local edits.