gen‑ai.news
← Back
Image

Black Forest Labs launches Flux 3 Image with multi-step editing that leaves the rest of your picture alone

Black Forest Labs launches Flux 3 Image with multi-step editing that leaves the rest of your picture alone

Black Forest Labs has launched Flux 3 Image, the image-focused component of its broader Flux 3 model family. The release builds on the company's existing Flux lineage with a set of capabilities aimed at giving users more precise, iterative control over generated images - areas where earlier generative models have often fallen short.

The headline feature is multi-step editing with localized changes. Users can make a series of targeted edits to an image, and the model is designed to leave unaffected areas visually consistent rather than allowing alterations to bleed into the broader composition. This kind of region-aware editing has been a persistent challenge in diffusion-based image models, where even small prompt changes can introduce unintended variations across an entire output.

Flux 3 Image also supports scene composition using bounding boxes, allowing users to specify where elements should appear within a frame. Up to ten reference images can be incorporated into a single generation, which opens the door to more structured workflows - useful for tasks like maintaining character consistency, matching a visual style, or assembling complex multi-element scenes. The model can output images at resolutions up to 4K, placing it in line with other high-resolution options currently available in the space.

Black Forest Labs has confirmed that open weights for Flux 3 Image are planned for release within the next few weeks. The company has previously made weights available for earlier Flux models, and that pattern of commercial access alongside open releases has helped establish Flux as a widely used base for fine-tuning and community development. How well Flux 3 Image performs in practice - particularly on the editing consistency front - will likely become clearer once independent testing begins following the open weights release.

Enjoy this story? Get the next one in your inbox.

Twice a week: the most important stories in generative image and video AI, distilled into a 2-minute read.

Free. Unsubscribe any time. No spam, ever.

Your next read

Apple is testing daily usage limits for more complex Siri AI interactions
Image

Apple is testing daily usage limits for more complex Siri AI interactions

Apple is testing daily usage limits on more advanced Siri and Apple Intelligence features in the latest developer betas of iOS, iPadOS, and macOS 27.3. Cloud-dependent tasks like image generation and multi-step personal requests would be capped, while basic Siri commands remain free and unlimited. The move signals a possible shift toward tiered access for computationally intensive AI features.

OpenAI wants to stop deepfakes from overwhelming child abuse tiplines
Image

OpenAI wants to stop deepfakes from overwhelming child abuse tiplines

OpenAI has announced a classifier designed to help child sexual abuse material investigators identify harmful imagery more efficiently, including content that is AI-generated. The tool will be made available free of charge to vetted organizations working in this space. The move comes as pressure mounts on AI developers to take more concrete steps toward safety.

You can now generate and edit GIFs with ChatGPT - here's how
Image

You can now generate and edit GIFs with ChatGPT - here's how

OpenAI has added GIF generation and editing to ChatGPT, giving users a way to create short animated images directly within the chat interface. The feature expands on the platform's existing image capabilities, extending them into the animated format. Here is what you need to know to get started.