Guide · Flux
The Best Flux Prompts
A curated, copy-ready library of high-leverage FLUX.1 prompts — built around how Pro, Dev and Schnell actually interpret subject, structure and style.
PromptInFlow AI Guides
General AI Models
Image & Video Models
AI Task Guides
- AI WritingRead guide
- AI MarketingRead guide
- AI CodingComing soon
- AI BusinessComing soon
- AI EducationComing soon
Flux is the open-weights image model that quietly became the production default for teams who need control. Where Midjourney optimizes for aesthetic and DALL·E optimizes for conversation, FLUX.1 — released by Black Forest Labs — optimizes for prompt adherence, photoreal output and accurate in-image text. That tradeoff matters: when you need a specific subject, in a specific pose, holding a specific label, with a specific brand color, Flux is often the right answer. The prompts in this guide are written for how Flux actually behaves — they're literal, structured and explicit about composition, because Flux rewards that.
This is a working reference, not a feature tour. Every prompt has been tested across the three flagship Flux variants — Pro for the highest fidelity, Dev for self-hosted and LoRA work, and Schnell for fast iteration — inside fal.ai, Replicate, ComfyUI and the Black Forest Labs API. You'll find prompts for photoreal portraits, product photography, brand imagery, in-image text and signage, character consistency with LoRAs, and ControlNet-style composition workflows.
Who is this guide for? Designers and marketers who need photoreal output with precise control. Developers shipping Flux via API for user-facing image generation. Operators running Flux Dev with custom LoRAs for character or brand consistency. Indie creators self-hosting in ComfyUI. If you've used Midjourney and thought "I need this exact subject in this exact pose with this exact text," Flux is the tool — and this guide is the prompt playbook.
To get the most from Flux, write the way you'd describe a photo to a literal-minded production assistant. Specify the subject, the action, the camera, the light, the environment and the desired output in plain language — Flux parses natural sentences far more cleanly than tag-soup prompts. Use Schnell to explore, Dev for self-hosted control and Pro for the hero shot. Bookmark this page — we refresh it whenever Black Forest Labs ships a new Flux variant or capability.
Editorial standards
How we selected these prompts
Why each prompt earns its place in the library.
Every prompt in this guide has been tested across the current Flux lineup and re-tuned until it produces a usable image without follow-ups. We test inside fal.ai, Replicate and ComfyUI and only ship prompts that hold up across variants, adapt cleanly to a new subject, and work with or without LoRAs depending on the use case.
Built for Flux's strengths
Each prompt leans into what Flux does best — prompt adherence, photoreal subjects, accurate in-image text, and clean composition control.
Natural-language structured
Flux parses full sentences better than comma-soup. Prompts read like a brief, not like a search query.
Variant-aware
We note when a prompt is tuned for Pro, Dev or Schnell — they have different sweet spots and we don't pretend they're interchangeable.
LoRA-friendly
Prompts that pair with LoRAs (for character or brand consistency) are flagged so you can run them in ComfyUI or via API with the right weights loaded.
Who this guide is for
This guide is built for anyone who needs Flux to deliver precise, photoreal, on-brief imagery — not just impressive defaults.
Designers
Brand imagery, product photography and editorial visuals with control over composition and color.
Developers
Shipping Flux via API in user-facing apps — image avatars, marketing assets, dynamic creative.
Marketers
Ad creative with accurate text-in-image, product shots and lifestyle imagery without a shoot.
Indie creators
Self-hosting Flux Dev in ComfyUI with custom LoRAs for character or world consistency.
Photographers
Generating references for shoots — lighting, location and composition studies that look real.
Product teams
Onboarding imagery, app store screenshots and in-product illustrations with exact subject control.
In this guide
What you'll learn
Skim the highlights, then jump to the section you need.
When Flux beats Midjourney (and when it doesn't)
Where Flux wins — control, photorealism, in-image text — and where Midjourney is still the better default.
35+ copy-ready Flux prompts by use case
Portraits, product, brand, signage, character series and composition-controlled workflows.
Pro vs Dev vs Schnell — which to use when
A clear rubric for picking the right Flux variant, with cost and quality tradeoffs.
Working with LoRAs and ControlNets
How to brief Flux Dev with character LoRAs and pose / depth control so the same subject appears across an entire series.
Framework
How great Flux prompts work
The six-part anatomy behind every prompt that consistently ships usable output.
Flux rewards literal, structured natural language. The prompts that consistently produce usable images share the same shape: S-S-C-L-T-O (Subject, Setting, Camera / medium, Light, Text-in-image, Output specifics). Skip a layer and Flux defaults to a competent but generic interpretation; specify each one and you get the exact image you described.
Subject
Open with a specific, concrete subject sentence: "A 40-year-old female architect in a charcoal wool coat, standing in front of a wall of blueprints." Flux parses subject-first prompts much more cleanly than vibe-first prompts.
Setting
Anchor the location with one or two visual details — "warm-lit Brooklyn studio with exposed brick," "sun-bleached coastal cliff at golden hour." Setting cues stop Flux from defaulting to generic studio backdrops.
Camera / medium
Name the camera and lens for photo ("shot on Sony A7 IV with an 85mm at f/1.8") or the medium for illustration ("matte digital painting, soft edges, limited palette"). This single layer is the biggest quality lever.
Light
Specify direction, quality and color: "soft warm window light from camera-left, gentle falloff into the background." Flux honours lighting cues precisely — use them.
Text-in-image
Flux is unusually good at rendering text. Wrap the exact text in quotes and name where it appears: 'the words "FRESH BAKERY" on a hanging wooden sign'. Keep text short (1–6 words) for best results.
Output specifics
Close with aspect ratio, resolution and any negative cues. Flux respects "no text, no watermark, no extra subjects" when you state them explicitly.
Three examples — and why they work
Photoreal commercial portrait
A 35-year-old female bakery owner in a flour-dusted navy apron, leaning against a wooden counter inside a warm-lit independent bakery with copper pans hanging in the background. Shot on a Sony A7 IV with an 85mm lens at f/1.8, soft warm window light from camera-left, shallow depth of field, magazine-cover quality, natural skin texture, no makeup retouching. 4:5 portrait crop, no text, no watermark.
In-image signage that actually reads
A handmade wooden hanging sign outside a small independent coffee shop, with the words "COLD BREW" carved into the wood in a clean block sans-serif and painted matte black. Mounted on weathered brick wall. Shot from a slight low-three-quarter angle, late afternoon warm sunlight grazing the sign from the right, shallow depth of field with the brick softly out of focus, photoreal commercial photography. 3:2 landscape, no other text in the frame.
Character series with LoRA
[Use with character LoRA: weight 0.85] A young space botanist in a worn olive flight suit kneels next to a glowing potted fern inside a softly-lit greenhouse module aboard a spaceship. Three-quarter side angle, soft cool overhead light with warm fill from the fern, photoreal cinematic still, anamorphic lens look, gentle film grain. 16:9 landscape, no text, no watermark.
Browse by use case
Featured prompt categories
Jump into the library that matches what you're working on — each category includes guidance on when to use it and what good output looks like.
- When to use
- Reach for Flux when you need control — exact subject, exact pose, accurate text, specific brand colors — or when you're shipping at scale and need consistent output across an API.
- What you can achieve
- Photoreal portraits and products, accurate signage and labels, character series via LoRAs, and composition-controlled images via ControlNets in ComfyUI.
- Best practices
- Write in full sentences, not tag-soup — Flux parses natural language better than any other open-weights model.
- Pick the variant deliberately: Schnell for exploration, Dev for self-hosting and LoRAs, Pro for hero shots.
- Always close with negative cues ("no text, no watermark, no extra props") for ad and product work.
- When to use
- Use Flux for ad creative that needs accurate brand text, product shots with exact label control, and lifestyle imagery generated at API scale.
- What you can achieve
- A/B-ready ad variants with on-brand text in the image, programmatic creative for paid social, and product imagery that doesn't need photoshop to fix garbled labels.
- Best practices
- Quote brand text exactly and name where it appears in the frame — Flux is the strongest model at rendering it.
- Match aspect ratio to placement at generation, not via post-crop, to preserve resolution.
- Use Flux Pro for the hero shot of a campaign and Schnell for the dozens of variants you'll test underneath it.
- When to use
- Use Flux for concept art, worldbuilding and editorial illustration where you want photoreal or carefully-stylized output you can iterate on with precise prompts.
- What you can achieve
- Mood boards with specific subjects (not just vibes), character continuity via LoRAs, and a consistent visual language across a series.
- Best practices
- For style consistency, train a small LoRA on a handful of reference images and run it on Flux Dev — it's the most reliable consistency lever in the open ecosystem.
- Combine Flux with ControlNets in ComfyUI for pose, depth and edge control when composition matters more than improvisation.
- Constrain palette explicitly; Flux honours "limited palette of olive, rust and cream" precisely.
- When to use
- Pair Flux with writing work for hero illustrations, ebook covers and blog imagery where you need a specific subject doing a specific thing.
- What you can achieve
- Cover art with accurate titles rendered in-image, editorial imagery consistent across an article series, and illustrations that match the article's exact concept.
- Best practices
- Mirror the article's tone in the prompt — calm palettes and soft light for serious pieces, bold contrast and saturated color for cultural ones.
- Render in-image titles directly with Flux when the design allows; only overlay text in post when the typography is fixed brand type.
- Generate the hero image first, then ask Flux for two companion images using the same setting and lighting language.
Hand-picked
Curated prompts
The prompts we reach for most often — copy, paste, adapt.
Editorial Food Photography
Top-down hero shot of a seasonal dish with soft natural light and a moody, magazine-quality palette.
Top-down editorial food photography of a roasted heritage carrot dish, hand-thrown ceramic plate, linen napkin, soft window light from the left, deep shadows, Kinfolk magazine palette --ar 4:5 --style raw
Veo Character Consistency Sheet
Keep the same character across multiple Veo shots with a reusable description block and shot list.
Character: 32-year-old woman, shoulder-length auburn hair, freckles, olive linen jacket. Reuse this exact description for every shot. Shot 1: medium close-up, golden hour. Shot 2: walking shot, overcast street. Shot 3: cafe interior, warm tungsten.
Cinematic Coffee Commercial
Slow-motion espresso pour with macro detail, warm morning light and a 35mm anamorphic film look.
Cinematic 5-second shot, slow-motion espresso pour into a ceramic cup, macro lens, warm golden morning light through a cafe window, shallow depth of field, anamorphic 35mm film grain, subtle steam.
At a glance
Quick reference tables
Cheat-sheets for picking the right model and the right prompt style.
Flux variants at a glance
Same family, different sweet spots.
| Variant | Best for | Watch out for |
|---|---|---|
| FLUX.1 Pro | Highest-fidelity hero shots, commercial work, API production | Most expensive per image — overkill for early exploration |
| FLUX.1 Dev | Self-hosted, LoRA / ControlNet workflows, indie + research | Slower locally; needs a capable GPU or a hosted endpoint |
| FLUX.1 Schnell | Fast iteration, drafting, real-time / interactive UX | Lower fidelity than Pro; not the right final-asset choice |
Flux vs Midjourney vs DALL·E
Where each model wins. Pick the right tool for the job.
| Task | Pick Flux when | Pick Midjourney / DALL·E when |
|---|---|---|
| Aesthetic-first creative | You need an exact subject and pose | You want the default to look beautiful (Midjourney) |
| In-image text | Accuracy matters — Flux is the strongest at this | Text is decorative and 1–2 words (Midjourney v7 is OK) |
| Character / brand consistency | You can train and run a LoRA on Dev | Quick consistency via --cref / --sref (Midjourney) |
| API and programmatic scale | Shipping in a product or pipeline | DALL·E inside ChatGPT for conversational generation |
| Self-hosting / privacy | Sensitive data, on-prem, or research | Hosted-only (Midjourney, DALL·E) |
Where to run Flux
Pick the surface that matches your stage.
| Surface | Best for | Notes |
|---|---|---|
| fal.ai | Fast hosted Pro / Dev / Schnell with low setup | Great for prototyping and production APIs |
| Replicate | Hosted Flux with a familiar API | Easy to swap variants and try LoRAs |
| Black Forest Labs API | Official Pro endpoint with the latest features | Use for hero-quality production |
| ComfyUI / local | LoRAs, ControlNets, full pipeline control | Best for advanced workflows on Flux Dev |
Method
How to write better prompts
Four habits that lift the quality of every prompt you write.
Write in sentences, not tags
Flux parses natural language more cleanly than any other open model. "A 40-year-old chef plating a dish under warm overhead light" outperforms "chef, plating, warm light, 85mm, photoreal, masterpiece."
Pick the variant deliberately
Schnell for the first 20 explorations, Dev when you need LoRAs or self-hosting, Pro only for the hero image you'll actually ship. Defaulting to Pro is the most common cost mistake.
Lean on in-image text
Flux's text accuracy is genuinely category-leading. If your design needs a sign, label, book cover or t-shirt graphic with short text, render it in Flux and avoid the post-production overlay.
Use LoRAs for consistency
For character, brand or world consistency across a series, train a small LoRA on 10–20 reference images and run it on Flux Dev. It's far more reliable than reference-image flags.
Avoid
Common mistakes
The patterns that quietly tank output quality.
- Writing tag-soup prompts inherited from Midjourney or SD habits — Flux ignores most of the noise.
- Defaulting to Flux Pro for every test image, then being surprised by the bill.
- Asking for long sentences of text in-image — Flux is great at short labels and weaker at paragraphs.
- Skipping the lens or medium line; without it, Flux produces a competent but generic image.
- Trying to do LoRA work on hosted Pro endpoints that don't support custom weights — use Dev for LoRAs.
- Forgetting negative cues on commercial work and getting extra props, extra subjects or unwanted text in the frame.
FAQ
Frequently asked questions
Quick answers about this guide and the prompts in it.
Keep reading
Related guides
The Best Midjourney Prompts
A curated, copy-ready library of high-leverage Midjourney prompts — built around how v6 and v7 actually interpret subject, style and parameters.
The Best Veo Prompts
A curated, copy-ready library of high-leverage Google Veo prompts — built around how Veo 2 and Veo 3 actually interpret subject, camera and motion.
The Best Kling Prompts
A curated, copy-ready library of high-leverage Kling prompts — built around Kling's strengths in cinematic motion, character consistency and longer takes.
The Best AI Marketing Prompts
A curated, model-aware library of marketing prompts that ship — built for the channels and tasks marketers actually run every week.
Explore
Related categories
Compatible models
Works well with
Fresh
Latest prompts
Recently added to the Prompt InFlow library.
Editorial Food Photography
Top-down hero shot of a seasonal dish with soft natural light and a moody, magazine-quality palette.
Top-down editorial food photography of a roasted heritage carrot dish, hand-thrown ceramic plate, linen napkin, soft window light from the left, deep shadows, Kinfolk magazine palette --ar 4:5 --style raw
Veo Character Consistency Sheet
Keep the same character across multiple Veo shots with a reusable description block and shot list.
Character: 32-year-old woman, shoulder-length auburn hair, freckles, olive linen jacket. Reuse this exact description for every shot. Shot 1: medium close-up, golden hour. Shot 2: walking shot, overcast street. Shot 3: cafe interior, warm tungsten.
Cinematic Coffee Commercial
Slow-motion espresso pour with macro detail, warm morning light and a 35mm anamorphic film look.
Cinematic 5-second shot, slow-motion espresso pour into a ceramic cup, macro lens, warm golden morning light through a cafe window, shallow depth of field, anamorphic 35mm film grain, subtle steam.
SEO Topic Cluster Builder
Generate a pillar page plus 12 supporting article ideas with target keywords and search intent.
Act as an SEO strategist. For the seed topic {topic}, produce one pillar page idea and 12 supporting articles. For each: working title, primary keyword, search intent, and one internal link target.
In summary
Conclusion
The ideas to carry forward — and where to go next.
Flux is the model that rewards intent. Subject, setting, camera, light, text and output specifics — name each layer in natural language and you'll consistently get the exact image you described, not a competent approximation. That control is why Flux quietly became the production default for teams shipping image generation in real products.
Treat each prompt in this library as a brief. Swap the subject and setting, keep the structure, and pick the variant deliberately — Schnell to explore, Dev for LoRAs and self-hosting, Pro for the hero. Combine Flux with a LoRA for consistency and with Veo or Kling for image-to-video, and you have a complete production pipeline.
When you're ready to push further, browse our image-AI prompt library, jump into the prompt generators to build a Flux-ready brief in seconds, or explore the Midjourney, Veo and Kling guides to see how each model fits into a modern visual workflow.
Continue Learning
Prompt Engineering Learning Path
A guided reading order through every Prompt InFlow flagship guide — start with the model you use most, then layer in topical skills.
- 1
Best ChatGPT Prompts
50+ tested ChatGPT prompts and the R-G-C-C-O-E framework that makes them ship usable output.
Model guide - 2
Best Claude Prompts
Long-context reasoning, on-voice writing and XML-tagged briefs for Claude 3.5 and Claude 4.
Model guide - 3
Best Gemini Prompts
Grounded, multimodal and long-context prompts that lean into Gemini's real-time strengths.
Model guide - 4
Best AI Writing Prompts
Voice, structure and editing — the writing playbook across ChatGPT, Claude and Gemini.
Topical guide - 5
Best AI Marketing Prompts
SEO, content, social, email, ads and brand — model-matched prompts for every channel.
Topical guide
Build prompts faster with Prompt InFlow
Browse 500+ tested prompts across every major AI model — or use our generators to build your own in seconds.