Black Forest Labs releases Flux 3 Image with multi-step editing, bounding boxes, up to 10 reference images and 4K output. What HK creators need to know.
Black Forest Labs has released Flux 3 Image, the image generation side of its Flux 3 model family. Unlike previous Flux models focused purely on speed or video, Flux 3 Image brings multi-step in-place editing — the ability to change parts of an image without affecting the rest of the scene. For Hong Kong creators producing brand content, product shots, or social media assets, this changes how you iterate on AI-generated visuals.
What Makes Flux 3 Image Different
Flux 3 Image covers text-to-image, image-to-image, text rendering, and photorealism in a single model. The headline feature is multi-step editing: you can modify specific elements in an image (swap a product background, change a model's outfit, adjust lighting) while leaving everything else untouched. Earlier models required full regenerations or manual masking — Flux 3 Image handles this natively.
Bounding box composition is another addition. Instead of describing an entire scene in one prompt, you can define spatial regions — "a bottle here, a hand there, a soft gradient behind" — and the model arranges them into a cohesive composition. Combined with support for up to 10 reference images, this gives art directors and brand teams fine-grained control over visual output.
The model outputs at up to 4K resolution, putting it in the same quality tier as Ideogram 4.5 and Midjourney. A free demo is available on the Black Forest Labs website, and API pricing starts at competitive rates — currently running at 50 percent off through October 8 for new users.
Why This Matters for Hong Kong Creators
Flux models are already widely used in Hong Kong's creative workflow. Flux Schnell powers high-speed iteration for social media content, while FLUX 3 Video handles short-form ad clips. Flux 3 Image fills the gap: a production-grade image model that supports iterative refinement — exactly what agencies need when a client asks for "the same layout, but brighter, with the logo on the right."
The open-weight release planned in the coming weeks is especially relevant for local production houses. Running Flux 3 Image on your own infrastructure means no per-image API costs for high-volume campaigns, full control over data privacy, and the ability to fine-tune on brand assets.
Working with Flux 3 Image in Your Pipeline
Flux 3 Image works naturally with tools like Cooly Studio, where you can generate initial concepts, apply multi-step edits, then export directly to your production pipeline. The bounding box and reference image features are particularly useful for:
- Product photography: Compose a hero shot with bounding boxes, refine with reference images of existing brand assets - Social media campaigns: Iterate on a single base composition — swap backgrounds, adjust colour palettes, test variations — without regenerating from scratch - Film and video pre-vis: Generate consistent style frames that match the same scene layout
With open weights on the horizon, expect third-party integrations and local deployments to appear quickly across the Hong Kong AI ecosystem.
Frequently Asked Questions
Q: How does Flux 3 Image's multi-step editing work? A: You select an area in the image and describe the change — the model modifies only that region while keeping the rest of the scene intact. Multiple edits can be chained in sequence.
Q: When will open weights be available? A: Black Forest Labs has confirmed an open-weight release in the coming weeks. Commercial weight licensing is available immediately for teams that need self-hosted deployment.
Q: How does Flux 3 Image compare to the existing FLUX 3 Video? A: They are separate models from the same family. FLUX 3 Video generates 20-second video clips with audio; Flux 3 Image generates still images at up to 4K with advanced editing controls. Many workflows use both in sequence.
Q: Can I use Flux 3 Image in Cooly Studio? A: Yes. Cooly Studio supports image generation and editing workflows that integrate with Flux 3 Image's API, making it straightforward to include in your existing production pipeline.