Skip to content

Image

Flux Kontext Max

Image

Available on Ofrai — chat, images, and video in one place with shared credits.

37 creditsModel details

Image studio

Preview

No output yet

Enter a prompt and hit Generate. Results will appear here.

Open studio

Example prompts

Choose an example to load its prompt and recommended settings into the Flux Kontext Max creator above.

How it works

Create with Flux Kontext Max in three steps

  1. 01

    Choose your inputs

    Keep Flux Kontext Max selected, then add a prompt and any supported reference assets.

  2. 02

    Set the output

    Choose the aspect ratio, resolution, and available Image options before generating.

  3. 03

    Generate and refine

    Review the result, adjust the prompt or settings, and generate another version from the same place.

Flux Kontext Max is for work that starts with an image you already have. Everything else about this engine is downstream of that: it reads a supplied picture, keeps what you did not ask it to change, and applies the change you described.

That is a different job from generation, and it is a job a lot of production actually needs — restyling a shot that exists, adjusting a background that is almost right, updating a mockup so the product inside it is the new one, running the same composition across a set of variations. Done by hand in a design tool, each of those is a small project. Done here, each is one instruction and one render.

What actually gets preserved

The value of an edit is entirely in what it leaves alone. Kontext treats the supplied image as the ground truth and treats your instruction as the only thing that is allowed to change, which means the subject, the pose, the composition, the lighting and the framing all survive unless you name them as targets. The practical test is whether you can make a change without describing the rest of the picture. If the answer is yes, this is the right tool and the instruction can be short.

If you find yourself re-describing the subject to be safe, you are fighting the engine rather than using it, and the result is usually a new image that happens to resemble the old one.

Instruction-based restyling

Restyle is the capability that gets bought for this engine and the one that is easiest to get wrong. The instruction has to describe the target look as a set of properties rather than as a mood: the period and medium of the lighting, the material of surfaces, the rendering style, the colour relationships. "Make it look like a 1970s photograph" is too vague to land reliably; "shot on colour film, warm tungsten light, soft halation around highlights, visible grain" gives the model things to move toward.

The general rule for this engine is that nouns and materials work and adjectives are weak, because a style is made of specific properties and the model needs to be given the properties.

Editing for a production change

The most valuable application is often the least creative one. A mockup exists with the old product in it and the new product needs to be in it. A scene needs the logo replaced. A background needs to be removed or replaced with a different one. A whole set of frames needs a consistent grade. In all of these the framing, the subject and the composition are settled and should not move, and the instruction is a single factual change.

This is the workflow where an edit engine is decisively cheaper than a design tool, and where regenerating from scratch is actively wrong, because regeneration throws away the thing you had already approved.

Where editing hits its limit

Two failures are common and both are worth recognising early. The first is asking for a structural change through an edit: moving a subject, changing the angle, adding a limb, turning a portrait into a landscape. The engine was not built to re-stage a scene, and asking it to will produce something plausible and wrong. The second is asking for an edit that contradicts the supplied image, for instance changing the light direction while keeping everything else.

When a request fights the picture, the picture usually wins, and you get a partial edit. The rule is simple: edit what can be edited in place, and regenerate when the change is structural.

How many edits to make on one image

Treat a sequence of edits as a conversation with a single aim, not as a set of independent requests. Each edit starts from the current state, so edits accumulate. That is useful when the steps are related — restyle, then correct the lighting, then replace one object — because each step inherits the last. It is wasteful when the steps are unrelated, because the second edit is trying to work around a result you did not want rather than moving toward the one you did.

Two or three coordinated edits per image is the usual sweet spot; past that, the picture has drifted far enough from the original that a regeneration with a better brief is usually faster and cheaper.

Using this engine alongside the others

On this catalogue the editing engines and the generating engines are separate tools, and a production run usually uses both: generate the base frame on whichever engine suits the look, then edit it here — to restyle, to place the real product, to correct one thing that came out wrong. The pairing works because the edit step inherits a composition you already approved, and because a specific, factual edit is cheaper and more predictable than trying to get the generating engine to land that detail on the first attempt.

The deciding question is never which engine is better; it is whether this job needs a picture made or a picture adjusted.

FAQ

What does Flux Kontext Max keep from the image I upload?
Everything you did not name as a target: the subject, the pose, the composition, the lighting and the framing all survive. If you find yourself re-describing the subject to be safe, the instruction is fighting the engine instead of using it.
Can I use it to restyle an existing photo?
Yes, and that is what it is built for. Describe the target as specific properties — the period and medium of the light, the rendering style, the material of surfaces — rather than as a mood. Nouns and materials land; bare adjectives are weak.
Should I edit an image or generate a new one?
Edit when the framing, subject and composition are settled and only one thing is wrong — restyling, a background change, a product swap, a grade. Generate when the change is structural, such as moving the subject or changing the angle, because this engine was not built to re-stage a scene.

FAQ

Can I use Flux Kontext Max directly on this page?
Yes. The creator above is the working Ofrai studio with Flux Kontext Max selected by default.
Where are my finished results saved?
Signed-in generations are kept in your Ofrai library so you can download or revisit them.
Which prompts work best with Flux Kontext Max?
Describe the subject, setting, visual style, lighting, and camera or motion clearly. The examples above provide practical starting points.
Can I edit the example prompts before generating?
Yes. Load any example, then change its wording, aspect ratio, references, or other supported settings in the creator.

Compare other Image models that can handle similar creative work to Flux Kontext Max.

Commercial use depends on the selected model and model terms.