AI Pre-Visualization: How to World-Build Your Next Sci-Fi Epic with Midjourney
AI Pre-Visualization: How to World-Build Your Next Sci-Fi Epic with Midjourney
Is AI coming for your job as a director, concept artist, or storyteller? Let’s get this out of the way: The answer is no. But a creative who knows how to direct AI will fundamentally redefine those roles for the better. As of July 9, 2025, the age of the lone genius is fading, replaced by the era of the ‘creative conductor’—the artist who can orchestrate a symphony of human and artificial intelligence. Forget the tired debates. Think of AI, specifically a tool like Midjourney, as the most powerful pre-visualization department you’ve ever had. It’s an engine for turning abstract ideas into tangible, visual worlds in minutes, not months.
Today, we’re not just generating a single cool image. We are going to conduct that symphony. We’ll design a cohesive visual language for a new science-fiction universe, from the sweeping vistas of its world down to the scuffs on a starship pilot’s boots. This is your lab session on moving beyond the one-off prompt and into systematic, AI-augmented world-building.
Phase 1: The Blueprint — Laying the Narrative Foundation with an AI Scribe
Every great visual world is built on a strong narrative foundation. Before we touch an image generator, we need a ‘design bible’—a document that outlines the lore, factions, technology, and aesthetic principles of our universe. A weak concept will lead to generic AI output. A strong one is the key to uniqueness.
Our creative co-pilot for this phase isn’t an image tool, but a large language model (LLM) like ChatGPT, Claude, or Llama 3. Its job is to be our tireless brainstorming partner.
Let’s start with a simple premise: ‘A civilization that fled a dying Earth now lives in symbiotic orbital habitats, bio-engineered from metallic asteroids.’
Instead of just asking for a story, we will be more strategic. We will prompt the LLM to act as a world-building consultant.
Strategist’s Log (LLM Prompting): When using an LLM for world-building, prompt it to take on a persona. Ask it to ‘Act as a senior world-builder for a AAA sci-fi game.’ This frames its responses, encouraging it to think in terms of visual and thematic consistency. Then, ask for specific, structured output, like lists of ‘visual motifs,’ ‘core materials,’ and ‘technological philosophies.’ This gives you a structured vocabulary to feed into your image prompts later.
Here’s how you might query your AI scribe:
“Acting as a senior world-builder, take this premise: ‘A civilization lives in symbiotic orbital habitats, bio-engineered from metallic asteroids.’ Give me a bulleted list of:
1. Three core visual motifs for this civilization (The ‘Archivists’).
2. The primary materials used in their architecture and clothing.
3. The design philosophy of their technology (e.g., is it slick and minimalist, or complex and organic?).
4. A name for a rival faction and their contrasting aesthetic.”
In a few seconds, the LLM might return something like this:
– Visual Motifs: Hexagonal lattices (representing honeycomb strength), flowing bioluminescent veins, fractal geometry.
– Materials: Iridescent chitin-steel, woven asteroid filaments, living crystal.
– Design Philosophy: Organic Synthesis. Technology is not built, but ‘grown.’ No hard edges, everything is curved and integrated.
– Rival Faction: The ‘Purifiers,’ who reject bio-integration. Their aesthetic is brutalist, utilitarian, and based on raw, sharp-angled asteroid iron.
Boom. In less than a minute, we have a rich, conflict-driven visual language to work with. Now, we can enter the image studio with a clear directive.
Phase 2: The Style Seed — Forging a Consistent Visual Signature
The biggest challenge for creators using AI is consistency. How do you generate a character and an environment that feel like they belong in the same universe? The secret is to first establish a “Style Seed”—a master image that defines the core aesthetic of your world.
We take the keywords from our LLM brainstorming session and translate them into the language of Midjourney. This first prompt is the most important one you will write.
The Prompting Studio: The Archivist World Seed
Go to the Midjourney Discord bot or web alpha. We’re building the home of our ‘Archivist’ civilization.
Copy and paste this prompt:
/imagine prompt: cinematic wide shot of the interior of a massive orbital habitat, bio-mechanical architecture in the style of Zaha Hadid and H.R. Giger, iridescent chitin-steel, walls covered in glowing fractal lattices and bioluminescent veins, grand and sacred atmosphere, volumetric lighting –ar 16:9 –stylize 250 –style raw
This prompt combines our keywords (‘chitin-steel’, ‘fractal lattices’) with artistic direction (‘Zaha Hadid and H.R. Giger’) and cinematic cues (‘volumetric lighting’). The parameters are critical.
You’ll get a grid of four images. Don’t just pick one and move on. Analyze them. Which one best captures the ‘Organic Synthesis’ philosophy? Which has the most compelling lighting? Select your hero image.
Strategist’s Log (Deconstructing the Parameters): The parameters are your camera controls. –ar 16:9 sets a widescreen, cinematic aspect ratio. –stylize 250 (a mid-range value) tells Midjourney to follow your artistic instructions closely but not go overboard into abstraction. Most importantly, –style raw often produces more photorealistic and less ‘opinionated’ or ‘house-style’ results, which is excellent for creating a unique look.
Once you’ve upscaled your chosen ‘Style Seed’ image, it becomes your anchor. It contains the DNA of your entire world. Now, let’s populate it.
Phase 3: The DNA Splice — Generating Consistent Characters and Props
This is where we move from amateur to pro. To ensure our character designs feel like they belong in our established world, we use a technique I call ‘DNA Splicing’. We will use our ‘Style Seed’ image as a direct input for our next generations. Midjourney’s Style Reference feature is built for this.
First, get the URL of your upscaled ‘Style Seed’ image. In Discord, you can right-click the image and select ‘Copy Link’.
Now, let’s design an ‘Archivist’ citizen. We will craft a new prompt for the character, but we’ll prepend it with our style reference command.
The Prompting Studio: The Archivist Pilot
This prompt uses a Style Reference (`–sref`) to make the character’s aesthetics match our world.
Copy and paste this prompt structure:
/imagine prompt: a female archivist pilot, thoughtful expression, wearing a flight suit made of woven asteroid filaments and living crystal, organic integrated helmet, full body character design sheet, neutral white background, studio lighting –sref [PASTE YOUR IMAGE URL HERE] –ar 2:3
The magic is in the –sref parameter. Midjourney will now analyze the colors, textures, shapes, and overall mood of your world’s image and apply that aesthetic DNA to the character.
The result is staggering. The pilot’s suit won’t be generic sci-fi; it will have the same iridescent, organic quality as the habitat’s walls. The technology will look ‘grown,’ not manufactured. You’ve created visual harmony across different subjects.
You can repeat this process for anything: vehicles, props, weapons, even the rival faction. For the ‘Purifiers,’ you’d first create their brutalist style seed, then use that image URL as the `–sref` for their characters. Now you have two distinct, visually consistent cultures.
The Big Questions: Your AI Debrief
“Is this workflow devaluing the craft of a concept artist?”
Absolutely not. It’s evolving the craft. A concept artist’s value is their imagination, taste, and problem-solving ability—not just their ability to render. This workflow liberates them from laborious initial sketching. Instead of spending a week on one polished painting of a single idea, they can generate, direct, and curate a dozen fully-formed visual directions. The artist becomes the Creative Director of the AI, guiding it toward a specific vision. The final work will still require human hands to refine, create orthographic views for 3D modelers, and add narrative details. This tool augments speed and exploration; it doesn’t replace artistry.
“What about copyright and using these assets in a commercial film?”
This is the most critical business question. As of mid-2025, the legal landscape is still solidifying. However, major AI services like Midjourney have clear terms: if you are on a paid plan (Pro or Mega), you generally own the assets you create. You can use them commercially. The caveat is that you cannot copyright the raw AI output in the same way you can a photograph you took. This means you may not be able to prevent someone else from generating a very similar image. For this reason, the key is the final, human-modified work. Your intellectual property lies in the unique world you’ve built, the story bible, the specific collection of curated images, and, most importantly, the final film, game, or product you create by integrating and altering these concepts. For any major production, consult with an intellectual property lawyer. Think of AI as providing you with royalty-free, but non-exclusive, concept assets.
“How can I ensure my world doesn’t look like someone else’s?”
Uniqueness comes from three places: 1. The Specificity of Your Initial Concept: A generic idea like ‘space marine’ will yield generic results. Our specific ‘Archivist’ concept is already more unique. 2. Your Prompt-Crafting and Curation: The unique combination of artists (Giger + Zaha Hadid), esoteric keywords (‘chitin-steel’), and your specific taste in selecting the style seed makes a huge difference. 3. The Post-AI Workflow: The most unique worlds use AI output as a foundation. Take these generated images into Photoshop or Procreate. Paint over them. Kitbash different AI elements together. Add custom decals and logos. Adjust color grading. The AI provides the clay; your human touch provides the soul and the defensible IP.
Your Creative Sandbox Assignment
Your mission, should you choose to accept it, is to design a single, hero prop for the ‘Purifier’ faction we brainstormed earlier. Their aesthetic is ‘brutalist, utilitarian, and based on raw, sharp-angled asteroid iron.’
- First, generate a ‘Style Seed’ image for the Purifier’s environment. Use keywords like `brutalist architecture, asteroid mine interior, harsh utilitarian lighting, raw iron walls, sharp angles, style of Ian McQue and Ridley Scott’s Alien –ar 16:9`.
- Upscale your favorite result and copy its image URL.
- Now, design a prop. Write a new prompt to create a ‘Purifier multi-tool.’ Use your ‘Style Seed’ URL with the `–sref` command. Describe the prop: `product design shot of a utilitarian multi-tool, welded raw iron, heavy and functional, no decorative elements, on a clean background –sref [PASTE YOUR PURIFIER URL]`.
- Look at the result. Does it feel like it was forged in the world you created? You’ve just performed a complete, stylistically consistent design workflow.
Your AI Integration Plan This Week
Don’t let this be a one-time experiment. Integrate this into your creative process with a simple schedule.
- Monday: Conceptualization (1 hour): Open a text document and a language model. Define the core premise of a project. Prompt the LLM for visual themes, material palettes, and cultural philosophies for two opposing factions.
- Wednesday: Style Seeding (1 hour): Jump into Midjourney. Spend the entire session just creating the ‘Style Seed’ images for your two factions. Don’t move on to characters. Just focus on creating the perfect, cinematic environmental shots that define your worlds. Experiment with different artist names and aspect ratios.
- Friday: Character & Prop Jam (1.5 hours): Using the `–sref` technique with your two style seeds, rapidly generate a lineup of 3 characters and 2 props for each faction. Don’t aim for perfection. Aim for volume. See how quickly you can create a diverse cast that feels thematically unified.
- Sunday: Review & Refine (30 mins): Create a simple mood board by pasting all your generated images into one canvas (you can use Figma, Miro, or even PowerPoint). Look at your world. What works? What doesn’t? What’s the next creative question you want to ask your AI co-pilot? This is the new creative cycle.
You are no longer just an artist staring at a blank page. You are a conductor, a director, an editor. You guide one of the most powerful creative engines ever conceived to bring your vision to life, at the speed of thought. Now go build some worlds.
Post Comment
You must be logged in to post a comment.