Use FLUX.1 Kontext to edit images with words

19 hours ago 1

FLUX.1 Kontext is a new image editing model from Black Forest Labs. It is the best in class model for editing images using text prompts, and the latest addition to the FLUX.1 family.

In our tests we’ve found Kontext to give accurate and brilliant results. It’s better and cheaper than OpenAI’s 4o/gpt-image-1 model (and there’s no yellow tint).

There are three models, two are available now, and a third open-weight version is coming soon:

FLUX.1 Kontext [pro]: State-of-the-art performance for image editing. High-quality outputs, great prompt following, and consistent results.
FLUX.1 Kontext [max]: A premium model that brings maximum performance, improved prompt adherence, and high-quality typography generation without compromise on speed.
Coming soon: FLUX.1 Kontext [dev]: An open-weight, guidance-distilled version of Kontext.

We’re so excited with what Kontext can do, we’ve created a collection of models on Replicate to give you ideas:

Multi-image kontext: Combine two images into one.
Portrait series: Generate a series of portraits from a single image
Change haircut: Change a person’s hair style and color
Iconic locations: Put yourself in front of famous landmarks
Professional headshot: Generate a professional headshot from any image

Run Kontext with an API

Like all models on Replicate, you can run Kontext with an API with just a few lines of code.

Here’s an example of how to run Kontext Pro with the Replicate Javascript client:

What you can do with Kontext

An image editing model lets you guide how an image changes. That means you can use text prompts to make small tweaks or big transformations while keeping the original image as your starting point. Change a car’s color, swap a background, or give a portrait a whole new style — all by telling Kontext what you want.

Kontext is really good with quick image edits. You’ll see in these examples that when you edit an image of a person, the person keeps their identity, whether the change is big or small.

For more complex changes, such as adding people or altering the setting, it’s best to describe each modification clearly. As long as your instructions per edit aren’t too complicated, being specific tends to give better results.

When editing people in images, small wording choices can make a big difference. A prompt like “transform the person into a Viking” might lead to a full identity swap. If you want to keep the same person, it helps to be more specific. You can focus your prompt on just the part you want to change, like clothing, eye color, or background.

Style transfer

When you’re prompting for style transfer, being specific makes all the difference (you’ll notice a theme with prompting FLUX Kontext). Instead of vague instructions like “make it artistic,” name the exact style, like “impressionist painting” or “watercolor sketch.” Referencing well-known movements or artists helps guide the model more clearly: think “Renaissance” or “1960s pop art.”

Kontext works quite well with reimagining images with different art media, for example.

If a style label doesn’t do the trick, describe the key traits that define it, such as “visible brushstrokes, thick paint texture, and rich color depth.” And if you want certain elements to stay the same, say so (“keep the original composition”). The more precise your language, the more intentional your results will be.

Text editing

Kontext can edit text directly in images, so you don’t need to recreate signs, posters, or labels from scratch. The clearest way to do this is often by quoting the exact text you want to change.

Here are some general tips when dealing with text:

Stick to readable fonts. Highly stylized text may not work as well.
Be explicit about what to keep. If preserving font style matters, make sure to mention it.
Match text length when possible. Big shifts in length can change the layout in ways you might not want.

Character consistency

Kontext is also great at keeping characters consistent, even through a bunch of edits. Start with a clear reference (like “the woman with short black hair”) and say what’s changing, whether it’s the setting, activity, or style. If you want the same person to stick around, just mention what to keep: face, expression, clothing, or whatever else matters.

Notice how we’re able to maintain character consistency through a long series of edits:

Occasionally, editing backgrounds and scenes while keeping characters consistent will require clarity to keep the subject in the same position, scale, or pose.

A short prompt can sometimes leave too much open to interpretation:

Put him on the beach

To better preserve a subject’s position, try a more descriptive prompt like this:

Change the background to a beach while keeping the person in the exact same position, maintain identical subject placement, camera angle, framing, and perspective. Only replace the environment around them.

Commercial use

If you generate images using Kontext directly on Replicate, you’re free to use those outputs in your commercial projects, including apps, marketing, or any other business use. The same goes for outputs from any fine-tuned versions of Kontext, as long as you’re running everything on Replicate.

It’s simple: run Kontext or its derivatives on Replicate, and you can use everything you generate, commercially.

Prompting tips and tricks

Kontext is surprisingly versatile. Whether you’re tweaking outfits, reimagining a photo as an oil painting, or swapping out signage text, the key is in how you write your prompts: clear, specific, and step-by-step when needed.

It’s easy to imagine FLUX Kontext powering some seriously creative apps. A visual story builder that evolves a character across scenes, a poster generator that updates brand content in seconds, or an AI-powered art canvas for rapid concept prototyping — there is so much potential. If you’re building something visual, Kontext is worth experimenting with.

Here’s a good TL;DR for prompting FLUX Kontext:

Be specific: Use clear, detailed language. Name exact colors, describe visual elements precisely, and choose direct action verbs. Avoid vague terms like “make it better.”
Start simple: Begin with basic changes. Test small edits first, then build on what works. Kontext supports iterative editing, so take advantage of that.
Preserve intentionally: Say what should stay the same. Use phrases like “while keeping the same facial features” or “maintain the original composition” to protect key elements.
Iterate when needed: Break complex edits into smaller steps. Big changes are easier to manage when done in sequence.
Name subjects directly: Use descriptive phrases like “the woman with short black hair” or “the red car.” Avoid pronouns—they’re often too vague.
Use quotation marks for text: Be exact when editing text. Writing “replace ‘x’ with ‘y’” works better than general instructions.
Control composition explicitly: When editing scenes, say if you want to keep things like camera angle or framing. It helps avoid unexpected layout changes.
Choose verbs carefully: Words like “transform” suggest a full rework. If you want more control, use specific actions like “change the clothes” or “replace the background.”