FLUX.1 Kontext tutorial: edit images, prompts & FLUX.2
Quick answer
To edit an image with FLUX.1 Kontext, upload a source image, describe one change, name the details that must stay fixed, and compare the result with the source. Use Pro for the lower-cost hosted route, Max when its output is worth the higher price on your own test, or Dev when local operation and its license fit. Black Forest Labs now labels Kontext previous generation and recommends FLUX.2 for new projects.
Need the current variant, price and license summary first? Read What is FLUX.1 Kontext? The facts below were checked September 13, 2026 against BFL's official documentation.
How to edit an image with FLUX.1 Kontext
Choose Pro or Max in BFL's API or playground, or run Dev under its applicable license and hardware requirements.
Upload the highest-quality source image you are allowed to use.
Write one instruction that names the change and the details that must remain fixed.
Generate one result, inspect it at full size, and save the source, prompt, model variant and output.
Apply the next edit only after the current result preserves the required identity, composition, text and product details.
Kontext can edit without a manually painted mask, but that does not guarantee that every untouched region will remain pixel-identical. Compare the result with the source rather than trusting the instruction alone.
A prompt structure that is easy to review
Use: change + target + desired result + preserved details. Keep the first request narrow.
Object change: Change the ceramic mug from red to cobalt blue. Keep its exact shape, white logo, handle position, table, lighting and camera angle unchanged.
Text edit: Replace “SUMMER” with “AUTUMN” on the poster. Preserve the font weight, alignment, paper texture, shadows and every other word.
Background change: Replace the office background with a bright kitchen. Keep the person’s face, hair, clothing, pose, crop and lighting direction consistent.
Style reference: Apply the reference image’s ink-and-watercolor style to this street scene. Preserve the buildings, perspective, people and sign wording.
Change the ceramic mug from red to cobalt blue. Keep its exact shape, white logo, handle position, table, lighting and camera angle unchanged.
Replace “SUMMER” with “AUTUMN” on the poster. Preserve the font weight, alignment, paper texture, shadows and every other word.
Replace the office background with a bright kitchen. Keep the person’s face, hair, clothing, pose, crop and lighting direction consistent.
Apply the reference image’s ink-and-watercolor style to this street scene. Preserve the buildings, perspective, people and sign wording.
Workflow recipe
Anime-style full-scene composition of the subject seated high in a large leafy tree, enjoying an ice cream cone, white t-shirt, and dark overalls. Tree adorned with colorful lanterns, overlooking a vibrant market town with multi-storied buildings, balconies, and lively crowds below. Low-angle perspective from below. Bright, warm sunlight with soft shadows, rich saturated colors—lush greens, warm yellows and oranges, with pops of red, blue, and pink. Whimsical, nostalgic, lively fantasy atmosphere with detailed brushwork and watercolor texture.
- Model
- nano-banana-2
- Format
- auto
BFL's Kontext prompting guide recommends direct instructions, explicit preservation language and quoted text for text replacement. The exact result remains probabilistic, so validate every required detail.
A fair five-image evaluation
A face or character where identity must remain stable.
A product with a logo, label and exact shape.
A poster requiring one precise text replacement.
A busy scene where one object must change without moving others.
A style-reference task that must preserve composition.
For each case, record variant, provider, prompt, resolution, attempts, charged amount and whether the result passes the written requirements. Compare cost per accepted image, not only price per request.
Try an instruction-based image edit
Upload one image, describe one controlled change, and inspect what stayed consistent before applying more edits.
Open AI Image EditorBrowser and model alternatives
Use Magic Hour AI Image Editor when you want a managed browser workflow and model choice rather than operating weights. The selected model and credit estimate can change, so verify the live option before assuming it is FLUX.1 Kontext. Our AI image editor comparison covers other editing workflows.
Frequently asked questions
State the change, identify the target, describe the desired result, and list the details that must remain fixed. Keep the first request narrow so you can see which instruction caused each change.
Quote the existing text and its replacement, then specify the typography, placement, material and surrounding details to preserve. Check every character in the output before using it.
Use the highest-quality source available, change one thing per prompt, name immutable details explicitly, and compare the whole output with the source at full size. A preservation instruction reduces ambiguity but does not guarantee pixel-identical untouched areas.
Start with one result and record the source, prompt, model, resolution and cost. Generate more candidates only when the first result misses a written requirement, then compare cost per accepted image rather than cost per request.
