Landscapes and environments are some of the most rewarding subjects in AI image generation — and some of the easiest to under-specify. This tutorial gives you a layered method for writing environments that feel like real places, whether the brief is a fantasy mountain range, a futuristic cityscape, or a clean studio backdrop.
An environment prompt has to do three jobs at once: describe a place, evoke a feeling, and (often) leave space for a subject to live inside it. Most beginners overshoot on the first job and skip the others, which produces postcard-perfect but emotionally flat images. We will fix that with a layered approach you can reuse on any environment brief.
Think of an environment as four nested layers: foreground, midground, background, and atmosphere. Foreground is what is closest to the camera — rocks, grass, a road. Midground is where the action sits — a cabin, a figure, a market stall. Background is the depth — distant mountains, a skyline, the horizon. Atmosphere is what fills the air between those layers — mist, dust motes, snowfall, smog, golden haze.
This four-layer thinking maps directly onto how diffusion models compose scenes. State all four and you get coherent depth. State only one and you get a flat, posterised image.
Useful keywords: rolling hills, alpine meadow, glacial lake, jagged dolomite peaks, dense pine forest, redwood grove, savanna grassland, lavender fields, terraced rice paddies, monsoon-soaked jungle, basalt sea cliffs, sand dunes at dusk. Time of day and weather are crucial: golden hour, blue hour, monsoon storm rolling in, light mist clinging to the valley floor, fresh snowfall, dry heat haze.
Useful keywords: narrow Lisbon alley with hanging laundry, neon-lit Shibuya crossing in the rain, brutalist concrete plaza, art-deco Manhattan rooftop, sprawling cyberpunk megacity at night, sleepy Italian village square at dawn, smoky Mumbai street food stall. Mention surfaces — wet asphalt, cobblestone, polished marble, weathered brick — because surfaces carry most of the urban atmosphere.
Useful keywords: floating sky islands above an endless sea of clouds, bioluminescent forest at midnight, vast desert with a half-buried colossal statue, towering crystal cavern, alien jungle with iridescent flora, underwater ruins lit by sunlight shafts, post-apocalyptic overgrown city, sleek Martian colony at sunrise. Anchor in a real-world reference for grounding: "in the style of Studio Ghibli landscape painting", "Moebius-inspired alien terrain".
Useful keywords: seamless paper backdrop in dusty pink, soft gradient background, minimal beige cyclorama, marble surface with soft window light, brushed concrete podium, polished wood texture, neutral grey studio backdrop, infinity curve. For product work, less is more — the background's job is to disappear behind the product.
Weak prompt — flat single-layer description
a beautiful mountain landscape
"Beautiful mountain landscape" hits the model's most generic mountain cluster — likely a postcard scene with snow-capped peaks, blue sky, and zero atmosphere. There is no foreground to anchor depth, no time-of-day signal, and no weather. The output will be technically pretty and entirely forgettable.
Strong prompt — four layers explicit
FOREGROUND: a stone shepherd's hut in the lower-left
third of the frame, a coil of rope hanging by its door,
wildflowers and dry grasses immediately around it.
MIDGROUND: a winding gravel path leading the eye into
the valley, a small herd of sheep grazing in the distance.
BACKGROUND: jagged Dolomite peaks catching the last
warm light of sunset, layers of receding ridges fading
into deeper blue.
ATMOSPHERE: thin alpine mist clinging to the valley
floor, dust motes catching the golden side-light,
clean cool air feel.
Style: photorealistic landscape photography, shot on
Hasselblad H6D with 50mm lens, deep depth of field,
Kodak Ektar 100 film aesthetic.
Mood: serene, vast, quietly spiritual.
--ar 16:9 --v 6 --style raw
The output now has real depth — your eye moves from the hut, along the path, past the sheep, up to the peaks, and out into the misty atmosphere. The image feels like a real place that exists, not a stock postcard. Subject and environment work together rather than competing.
f/11 if your tool responds to it.Tip: When the result feels flat, ask: "Which of the four layers is missing?" Nine times out of ten the answer is atmosphere. Adding a single line — "soft mist drifting through the valley" — transforms the whole image.
Pick a real place you know well — your local park, a beach you have visited, a street you grew up on. Write a four-layer prompt for it. Generate and compare to your memory of the place. Where did the model under-describe?
Generate the same landscape (rolling hills with a path) in four atmospheres: morning mist, harsh midday clarity, golden hour haze, and a rolling thunderstorm. Subject stays identical; only atmosphere changes. Notice how mood transforms.
Write a clean studio background prompt for a fictional skincare product. Describe the surface, the backdrop colour, the light direction, and the props (or lack of them). Keep the prompt under 30 words — for product backgrounds, restraint is the skill.
Sign in to join the discussion and post comments.
Sign inFoundations of Prompt Engineering
The must-know basics of prompt engineering. Learn what prompts are, how AI models read them, and how to write clear instructions that get great results.
Advanced Prompt Engineering Techniques
Master the powerful techniques AI experts use every day. Chain-of-thought, RAG, agents, function calling, prompt evaluation, and much more — 20 deep-dive tutorials.
Prompt Engineering for Business & Productivity
Use AI to work smarter — automate tasks, make better decisions, and communicate professionally. 12 practical business prompt tutorials for professionals.
Prompt Engineering Projects & Real-World Applications
Twelve hands-on projects that turn prompt engineering theory into a portfolio. Build chatbots, content generators, RAG systems, and more.
Prompt Engineering for Specific AI Tools
Tool-by-tool mastery — deep dives into ChatGPT, Claude, Gemini, GitHub Copilot, Midjourney, Stable Diffusion, and more. Learn the exact prompting techniques each platform rewards.
Prompt Engineering for Data Science & Analytics
Supercharge your data workflows with AI. 15 practical tutorials on using prompt engineering for data cleaning, EDA, machine learning, SQL, visualisation, and more.