Flux Prompt Creator

This guide is structured to take you from the core philosophy to advanced techniques, with plenty of examples. Think of it as your personal manual for becoming a "Flux Director."

Part 1: The Core Philosophy - You Are a Director, Not a Librarian Before we write a single word, it's crucial to understand the mindset for prompting Flux 1 Dev.

Old Way (Librarian): You give the AI a list of tags. cat, fluffy, sitting, window, sunbeam. The AI tries to find an image that matches these labels. Flux 1 Dev Way (Director): You describe a complete scene, a moment in time, with emotion, context, and artistic intent. You are not just listing objects; you are painting a picture with words. Flux 1 Dev has been trained on a massive dataset and has a deep, nuanced understanding of natural language. It excels when you talk to it like you would describe a photograph or a painting to another person. Be descriptive, be evocative, and be specific.

Part 2: The Anatomy of a Master Prompt - The Prompting Pyramid A great prompt isn't just a jumble of words. It's built in layers, like a pyramid. Each layer adds more control and detail.

The Pyramid Structure (from most important to least):

Subject: The main "who" or "what" of your image. Action & Context: What the subject is doing, and where/when it's happening. Artistic Medium & Style: How the image should look (e.g., photo, painting). Composition & Framing: How the scene is arranged and viewed (camera angle). Lighting & Atmosphere: The mood and time of day. Color & Palette: The dominant colors and their feel. Final Polish & Details: The technical specifics that add realism or artistic flair. Let's break down each layer in detail.

Part 3: Deep Dive into Each Layer of the Pyramid Layer 1: Subject (The "Who" or "What") This is the foundation. Be as specific as possible. Don't just say "a dog." Say "a scruffy, one-eyed terrier mix with a patch of brown fur over his left eye."

Vague: a woman Good: an old woman Excellent: a 70-year-old woman with deep laugh lines, silver hair tied in a messy bun, and kind, crinkled eyes Actionable Tip: Use sensory details. What does the subject look like, feel like, or sound like (implied)? a sleek, black motorcycle with chrome accents that glint in the light.

Layer 2: Action & Context This brings your subject to life. What is it doing? Where is it? This layer prevents static, boring images.

Vague: a man in a forest Good: a man is walking through a forest Excellent: a man in a worn leather jacket is kneeling by a moss-covered stone, examining strange glowing runes carved into it Actionable Tip: Combine subject and action into a single, descriptive sentence. This is where you start telling a story.

Layer 3: Artistic Medium & Style This is one of the most powerful layers. It tells Flux how to render the scene. Be explicit.

Photography Styles: vintage photograph from the 1970s, high-fashion editorial photo, National Geographic style wildlife photograph, long exposure shot of a waterfall, macro shot of a dewdrop, black and white street photography Artistic Mediums: oil painting with thick impasto brushstrokes, delicate watercolor illustration, detailed ink drawing, charcoal sketch, anime key visual, Studio Ghibli film still, Ukiyo-e woodblock print Digital Styles: hyperrealistic 3D render, pixel art, low poly game asset, cyberpunk interface design, glitch art Actionable Tip: Combine mediums for unique effects. An oil painting in the style of Van Gogh, but depicting a futuristic cityscape.

Layer 4: Composition & Framing This is your camera lens. You are the cinematographer. Use photographic terms to guide the AI.

Shot Types: extreme close-up, close-up, medium shot, full-body shot, wide shot, extreme wide shot / establishing shot Camera Angles: low-angle shot (makes subject look powerful), high-angle shot (makes subject look vulnerable), dutch angle (creates unease), bird's-eye view, worm's-eye view Composition Rules: rule of thirds composition, leading lines, symmetrical composition, dynamic asymmetry Example: Instead of just a knight, try a low-angle shot of a knight, making him look heroic and imposing against the sky.

Layer 5: Lighting & Atmosphere Lighting is everything for mood. It can turn a peaceful scene into a sinister one, or a simple scene into a dramatic one.

Natural Light: golden hour lighting, soft morning mist, harsh midday sun, dramatic sunset, blue hour, overcast day with soft, diffused light Artificial Light: cinematic lighting, dramatic backlighting, soft studio lighting, neon glow from a city sign, candlelit, moonlight Atmosphere: eerie and foggy, serene and peaceful, tense and suspenseful, joyful and vibrant, melancholy and somber Actionable Tip: Describe the source and quality of the light. The scene is lit by the warm, flickering glow of a campfire, casting long, dancing shadows.

Layer 6: Color & Palette Direct the color scheme to evoke specific emotions.

General Descriptors: vibrant and saturated colors, muted and desaturated palette, monochromatic scheme, pastel colors, earthy tones Specific Colors: dominated by shades of deep blue and purple, a splash of bright red in an otherwise monochrome scene, warm autumn color palette of orange, red, and yellow Actionable Tip: Link color to mood. A desaturated, cold color palette to create a feeling of isolation.

Layer 7: Final Polish & Details These are the technical modifiers that add the final layer of quality and realism.

Detail Level: hyperrealistic, hyperdetailed, intricate details, sharp focus Camera Effects: depth of field, shallow depth of field, bokeh, motion blur, lens flare, film grain Engine/Renderer: (For digital styles) Octane render, Unreal Engine 5, V-Ray render Part 4: Putting It All Together - Worked Examples Let's build some prompts from simple to complex using the pyramid.

Example 1: A Simple Portrait

Level 0 (Vague): a photo of a wizard Level 5 (Good): An old wizard with a long white beard and a pointed hat, oil painting. Level 7 (Excellent - The Full Prompt): A detailed oil painting of an old wizard with a long, flowing white beard and piercing blue eyes. He is sitting in an ancient library, surrounded by towering bookshelves filled with leather-bound tomes. A single, glowing orb floats above his open hand, illuminating his face with a magical blue light. The scene is a medium shot, with a shallow depth of field that blurs the background books. The atmosphere is mysterious and scholarly. Intricate details on the texture of the books and the fabric of his robes. Example 2: A Landscape

Level 0 (Vague): a forest Level 5 (Good): A fantasy forest with glowing mushrooms, digital art. Level 7 (Excellent - The Full Prompt): A wide shot, breathtaking fantasy landscape at twilight. An ancient, moss-covered forest with bioluminescent mushrooms and glowing flora casting an ethereal blue and purple light on the scene. A crystal-clear river winds through the forest, reflecting the star-filled sky above. The atmosphere is magical and serene. Cinematic lighting, hyperdetailed, vibrant colors, rendered in Octane, sharp focus. Part 5: Advanced Techniques & Common Pitfalls Advanced Techniques Iterative Prompting (The Conversation Method): Don't try to get the perfect image on the first try. Start with a core idea, generate an image, see what you like and don't like, then refine your prompt. Step 1: a robot in a city -> Result: A generic robot. Step 2: a weathered robot sitting on a rooftop, looking at the rain -> Result: Better, but the city is boring. Step 3: A weathered robot with exposed wiring, sitting on the edge of a gritty cyberpunk rooftop, looking down at the rain-slicked streets below, neon signs reflecting in the puddles, cinematic lighting -> Result: The one. Negative Prompts (--no): Use this to remove specific, unwanted elements. Flux is very good at following positive prompts, so use negative prompts sparingly. Good use: --no text, words, watermark, signature Less effective use: Trying to define the whole image with what you don't want. It's better to be clear about what you do want. Prompt Weighting: While Flux is less reliant on syntax weighting than older models, you can still use it to add emphasis if a specific element is being ignored. The common syntax is (word:1.3) to increase weight and (word:0.7) to decrease it. Example: A woman in a (red dress:1.3) standing in a crowd of people wearing gray clothes. This tells Flux to make the red dress a dominant feature. Common Pitfalls and How to Fix Them Problem: "Melted" or anatomically incorrect subjects (e.g., six-fingered hands). Cause: The prompt is too complex or has conflicting actions. Fix: Simplify. Focus on one primary action. Use terms like correct anatomy, detailed hands. Sometimes, generating a full-body shot and then cropping is easier than getting a perfect hand close-up. Problem: The AI ignores parts of your long prompt. Cause: The most important words might be buried at the end. Fix: Reorder your prompt. Put the most critical subject and action at the very beginning. The first few words often carry the most weight. Problem: The image looks generic or "AI-like." Cause: The prompt is too generic. Fix: Add unique, specific details! Instead of a car, try a 1965 Ford Mustang with a dent in the fender and a "For Sale" sign in the window. Uniqueness in your prompt leads to uniqueness in your image. Part 6: The Final Workflow & Mindset Checklist Envision: Have a clear idea in your head. What's the story? What's the mood? Build (Pyramid): Write your prompt layer by layer, starting with the subject. Generate: Create your first set of images. Analyze: What worked? What didn't? Is the composition right? Is the mood there? Refine: Edit your prompt. Add, remove, or rephrase words based on your analysis. Be specific. Repeat: Go back to Step 3. Prompting is an iterative process of discovery.