Black Forest Labs In-Context Editor
Swap an outfit, a season, or a single prop while Flux Kontext Pro holds your subject, lighting, and composition exactly where the original photo put them.
Your generated image will appear here.
Enter a prompt and click 'Generate with Flux Kontext' — or start from an example:
Examples
Drag the handle to compare each source and result, or open the viewer for a closer look.
consistent changes / portrait workflows
Prompt: Change the charcoal blazer to a camel wool coat over the same white shirt, and move the woman from the office into a bright co-working lounge with green plants and warm wood shelves softly blurred behind her. Keep her face, hairstyle, skin tone, smile, and the natural window light direction exactly the same.
consistent changes
Prompt: Change the LED ring inside the transparent pebble speaker from warm orange to cool ice blue, and replace the walnut desk with a polished black marble surface with a subtle reflection. Keep the speaker's pebble shape, transparency, internal drivers, studio lighting, and light gray background exactly as they are.
prompt edits
Prompt: Change this rice terrace landscape from spring sunrise to golden autumn harvest: ripe amber rice stalks on the terraces, warmer low sunlight, the mist slightly thinner, the farmer's clothing a touch heavier. Keep the terrace geometry, the water reflections, the farmer's position, and the panoramic composition exactly the same.
prompt edits on text
Prompt: Change the artist name at the top from "MIRA VALE" to "NOVA REY" in the same thin white uppercase letter style, and change the album title at the bottom from "saltwater hymns" to "glass currents" in the same small lowercase italics. Keep the underwater swimmer, the golden surface light, the silver fish, and the deep teal palette completely unchanged.
prompt edits
Prompt: Recolor this flat illustration from the warm terracotta-and-cream palette to a cool winter palette: slate blue mug interior, soft ice-blue steam clouds, a cream and pale gray background. Keep the person, the laptop, the bookshelf and lamp inside the mug, the composition, and the negative space on the right exactly the same.
fast Flux workflows
Prompt: Turn the drink in the mug into a matcha latte with a simple leaf-shaped latte-art pattern, and add soft morning sunlight falling across the counter from the left with a gentle highlight on the mug. Keep the mug's shape, color, position, the wooden counter, and everything else in the kitchen exactly the same.
Source: RandalWhitmore — CC-BY 2.0 via Flickr.
Flux Kontext Pro is Black Forest Labs' in-context image editing model: instead of regenerating a picture from scratch, it reads your uploaded photo as the scene to preserve and applies only the change your prompt describes. That is why a request like "turn the grass into snow" leaves the person, the fence, and the late-afternoon light alone.
On RenderFlow AI, the dedicated Flux Kontext canvas exposes the controls that matter for this way of working — a single image upload, a Guidance Scale from 1 to 20, and nine aspect-ratio presets — at 4 credits per edit, with Flux Kontext Max one step up when a final frame needs sharper type and texture.
Start from the strongest original you have, because Kontext Pro preserves everything your instruction does not name. Opening from this page drops you into the Flux Kontext canvas with the model preselected — one image slot, no masking or layers to prepare.
Write the edit as a single, concrete instruction — "change the jacket to red leather" or "make the scene winter" — and quote any words that must appear. Stacking three requests into one prompt is the most common way to pull in details you wanted untouched; queue the next change as a follow-up pass instead.
The default 3.5 balances fidelity against change for most retouch work. Nudge the 1–20 slider upward when the edit barely registers, and pull it back down when textures or faces start to look over-processed.
The canvas offers nine presets from 21:9 to 9:21. Choose the crop the asset will actually ship in — 16:9 for a web hero, 9:16 for stories, 1:1 for catalog slots — so the edit composes for its destination instead of being cropped afterward.
If the frame needs crisp in-image text, or it is a hero asset where every texture will be inspected, rerun the winning prompt on Flux Kontext Max at 8 credits. For everything else, Pro's cheaper pass is the right finishing tool.
Single, well-scoped changes to an existing photo: object and prop swaps, outfit changes, season or weather shifts, background replacement, and style transfers that must keep the original composition. It is an editor, not a generator — for images created from scratch, use Flux Pro.
Both apply in-context, natural-language edits, but Max is the premium tier at 8 credits per image versus Pro at 4, with stronger typography rendering and maximum detail fidelity. Draft and iterate on Pro, then move to Max for final hero frames and edits that involve legible in-image text.
Guidance Scale (1–20, default 3.5) controls how aggressively the model follows your instruction versus staying glued to the original image. Most edits work between 3 and 7; push higher only when the change is not showing up, and lower it if the result starts to look synthetic.
Yes — identity preservation is its signature strength. Upload the portrait, describe what changes around the person, and chain follow-up edits; the face and identifying features stay consistent across the series.
The usual causes are an instruction that is too broad ("make it better"), several changes packed into one prompt, or a Guidance Scale set too high. Narrow the instruction to one concrete change, name what should stay, and return Guidance Scale toward 3.5.
Nine presets are available in the canvas — 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, and 9:21 — so you can compose the edit for its final placement instead of cropping afterward.
Use the premium FLUX Kontext image editing model.
Generate professional images with Flux 1.1 Pro.
Edit images with ByteDance Seedream V4 Edit.
Use advanced Alibaba Qwen image editing for product and layout changes.
Edit photos with ByteDance SeedEdit V3 prompt-based image editing.
Generate multi-image visual sequences with ByteDance Seedream V4.