Chrome ExtensionFLUX.2 Image Prompt Generator
Generate optimized prompts for all FLUX.2 variants — Klein (fastest), Dev (32B open-weight), Pro (production), Flex (max precision), Max (best edits), and Kontext (editing). Up to 10 reference images, 4MP native output, 32K-token prompts, and 92% text-rendering accuracy. Built around the official Subject + Action + Style + Context framework with positive-only phrasing.
Be specific: describe who/what is in the image with details
Quick Palettes
Generated Prompt
Fill in the form and click "Generate" to create an optimized FLUX.2 prompt.
Tip: Be as specific as possible with your subject description. Instead of "a dog," try "a golden retriever puppy with a red bandana."
FLUX.2 Tips
- • FLUX.2 has a single prompt field — NO negative prompts. Use POSITIVE phrasing ("sharp focus" not "no blur")
- • Front-load the subject. Subject + Action + Style + Context. Sweet spot: 30-80 words (up to 32K tokens supported)
- • Text rendering hits ~92% accuracy with text in "quotes", kept short (1-4 words), placed on clear surfaces (signs, covers)
- • Pass HEX codes through directly — "#FF6B35 accent" works as a color directive in FLUX.2
- • Pick the variant: klein for sub-second drafts, dev for open-source, pro for production speed, flex for precision typography, max for edits, kontext for image editing
- • Reference up to 10 images for character/product/style consistency — assign each a clear job
- • Name camera/lens/film stock to control color science and grain (Kodak Portra 400, Cinestill 800T, Hasselblad H6D)
FLUX.2 Prompt Templates
Copy-ready image briefs built around what this model actually rewards. Swap the [BRACKETED] parts for your own subject and settings, then paste straight into the tool.
Photoreal Scene
PhotographyFlowing prose with the subject front-loaded — FLUX.2 prioritises what comes first.
A [SUBJECT] [DOING WHAT] in [LOCATION]. Shot on a [FOCAL LENGTH] lens at [APERTURE], [LIGHTING, e.g. overcast daylight through a north-facing window]. The [MATERIAL DETAILS, e.g. brushed steel surface and worn oak grain] read clearly. Foreground: [ELEMENT]. Mid-ground: [SUBJECT]. Background: [ELEMENT] falling away softly. Tack-sharp focus on the subject, natural colour, real-world physics.
Typography Poster
TypographyFLUX.2 hits ~92% text accuracy — keep the words short and on a clear surface.
A [STYLE] poster. Large heading reading "[1-4 WORDS]" in bold [FONT STYLE, e.g. condensed sans-serif], [TEXT COLOUR], across the upper third. Below it, [SUBJECT/ILLUSTRATION] centred. [BACKGROUND DESCRIPTION]. Flat even lighting, high contrast between text and background, generous margins, clean print-ready composition.
Character Reference Sheet
CharacterConsistency across views, using FLUX.2's multi-reference capability.
Character reference sheet for [CHARACTER — build, age, hair, defining features], wearing [OUTFIT WITH MATERIALS]. Three views in a single row: front, three-quarter, profile. Identical proportions, identical palette, identical costume detail in every view. Plain [COLOUR] background, even flat lighting, no shadows, full body with even margins around each view.
Product on White
ProductE-commerce imagery where positive phrasing replaces negative prompts.
A [PRODUCT] centred on a seamless white background, three-quarter view, filling roughly 70% of the frame. [MATERIAL AND FINISH DETAIL]. Flat even studio lighting with a soft contact shadow directly beneath. Clean uncluttered background, single clear subject, tack-sharp focus edge to edge, true-to-life colour, no colour cast.
Architectural Visualisation
ArchitectureLayered foreground/mid/background description, FLUX.2's preferred structure.
[BUILDING TYPE] in [SETTING], [ARCHITECTURAL STYLE], [PRIMARY MATERIALS]. Foreground: [LANDSCAPING OR STREET ELEMENT]. Mid-ground: the building at three-quarter angle showing two facades. Background: [CONTEXT] under [SKY CONDITION]. [TIME OF DAY] light raking across the facade, accurate shadows, real-world lighting physics, architectural photography.
Editing with Kontext
EditingSingle targeted change with an explicit preservation clause.
Change the [ELEMENT] to [NEW STATE, e.g. weathered copper with green patina]. Keep the composition, the lighting direction, the shadows, the background and every other element exactly as they are. Match the new material's response to the existing light.
How to Use the Prompt Generator
Pick Variant & Describe Your Vision
Choose your FLUX.2 variant — Klein for sub-second drafts, Pro for production, Flex for precision typography, Max for edits, Kontext for image editing. Pick a use case category, set resolution and color palette, and describe your scene with subject + action + style + context. Front-load the subject.
Generate Your Prompt
Click Generate. The tool builds a 30-80 word brief in flowing prose, quotes any literal text for crisp rendering, passes HEX color codes through directly, names camera/lens/film stock for color science, and converts any exclusions into positive phrasing (FLUX.2 has no negative prompt field).
Copy & Create
Paste into the BFL Playground, BFL API, Replicate, fal.ai, getimg.ai, Cloudflare Workers AI, ComfyUI, or Krea. For typography work, render at 4MP. For multi-reference workflows, upload up to 10 images and reference each by index.
Frequently Asked Questions
What is FLUX.2?
FLUX.2 is Black Forest Labs' next-generation image model family, launched in 2026 to challenge Nano Banana 2, GPT Image 2, and Midjourney. It introduces five variants — pro, flex, dev (32B open-weight), max, and klein (sub-second on consumer hardware) — plus the FLUX Kontext line for image editing. Headline upgrades over FLUX.1: up to 10 reference images per request, 4MP native output, 32K text input tokens, frontier-grade photorealism with real-world lighting and physics, and reliable complex typography for production typography, infographics, memes, and UI mockups.
How does this prompt generator work?
You provide structured inputs — scene, subject, style, composition, lighting, color palette, FLUX.2 variant, generation mode, and any preserve list. Our AI assembles them into a FLUX.2-optimized prompt: Subject + Action + Style + Context structure, front-loaded subject, positive phrasing only (no negatives), quoted short text for typography, and HEX color codes passed through directly.
What's the difference between FLUX.2 variants?
FLUX.2 [klein] — fastest, sub-second generation, Apache 2.0 open-source, runs on consumer hardware. FLUX.2 [dev] — 32B open-weight model on Hugging Face, frontier performance, deploy yourself. FLUX.2 [pro] — production-ready API, state-of-the-art quality at max speed. FLUX.2 [flex] — maximum precision and text rendering, full creative control. FLUX.2 [max] — best editing consistency across tasks. FLUX Kontext — separate editing-focused model for advanced reasoning, style transfer, and identity-locked edits.
How accurate is text rendering in FLUX.2?
FLUX.2 [pro] hits roughly 92% accuracy on complex text layouts and is one of the strongest text-rendering models available. To maximize accuracy: enclose literal text in "quotes," keep it short (1-4 words), place it on clear surfaces (signs, book covers, storefronts, T-shirts), and avoid lengthy sentences or tiny text. For typography-heavy work, pick the [flex] variant.
Why no negative prompts?
FLUX.2 uses a single prompt field — there is no separate negative prompt input. The fix is positive phrasing: instead of "no blur, no distortion, no artifacts," write "tack-sharp focus, clean lines, smooth surfaces." Our generator automatically converts your exclusion list into positive phrasing that FLUX.2 actually understands.
How does multi-reference work?
FLUX.2 accepts up to 10 reference images simultaneously with the best-in-class character, product, and style consistency. For best results, assign each reference a clear role in the prompt: "@Image1 defines the subject. @Image2 defines the style. @Image3 defines the lighting." This works across all variants but is strongest in [pro] and [max].
What resolutions and aspect ratios does FLUX.2 support?
FLUX.2 generates and edits up to 4MP (megapixels) natively while preserving detail and coherence. Any aspect ratio is supported: 1:1, 4:3, 3:4, 4:5, 16:9, 9:16, 21:9, 2:3, 3:2, plus custom dimensions. Output speed scales — [klein] runs sub-second, [pro]/[flex] generate in under 10 seconds.
How long should my prompt be?
FLUX.2 accepts up to 32,000 tokens, but the sweet spot is 30-80 words for most images. Short prompts (10-30 words) work for simple subjects — the model auto-fills details. Long prompts (80+ words) are for complex multi-element scenes with purposeful detail. Our generator targets the 30-80 word range by default and expands when you describe a layered or multi-shot composition.
Can I use HEX color codes?
Yes — FLUX.2 understands HEX codes directly. Your color palette picker passes hex codes through to the prompt as concrete directives ("#FF6B35 as the dominant accent, #0F172A for grounding"). This is the most precise way to control color in FLUX.2.
Is this tool free to use?
Yes. You get 1 free prompt generation per day with no signup required. For unlimited access, sign up for a Promptslove membership which includes all AI tools and 20,000+ premium prompts.
Where can I use the generated prompts?
The prompts are tuned for FLUX.2 across the BFL Playground and BFL API, Replicate, fal.ai, getimg.ai, Cloudflare Workers AI, ComfyUI, and Krea. The [dev] variant is a 32B open-weight model on Hugging Face you can self-host. Prompts also produce strong results on Nano Banana 2, GPT Image 2, Seedream 5, and Midjourney.
Want Unlimited AI Prompt Generation?
Get unlimited access to all AI tools, 20,000+ premium prompts, courses, and resources designed to maximize your creative output.