What Is an AI Image Prompt?
An AI image prompt is a text description that tells an AI model (like Gemini, Imagen, or DALL-E) exactly what image to generate. The quality of the prompt directly determines the quality of the result.
Most people write prompts as a single messy sentence. Professionals use a structured approach — breaking the prompt into distinct components that the AI can interpret precisely.
Key insight: Modern AI models like Gemini Nano Banana are extremely sensitive to prompt structure. Small changes in wording can completely transform the output. A structured prompt gives you control over every aspect of the image.
The 7-Part AI Prompt Structure
D.O.M.A. Image Studio uses a proven 7-field prompt system. Instead of writing one long sentence, you define each visual element separately. This makes prompts reusable, testable, and improvable.
| Field | What It Controls | Example |
|---|---|---|
| Subject | The main element in the image | A woman in a vintage dress |
| Action | What the subject is doing | Sitting in a café, drinking coffee |
| Environment | Location, background, surroundings | Cozy jazz club with dim blue lighting |
| Style | Artistic or visual style | Cinematic, film noir, dramatic |
| Lighting | Light quality and direction | Dramatic chiaroscuro lighting |
| Camera | Lens, angle, depth of field | DSLR, 85mm f/1.2, shallow depth of field |
| Emotion | Mood and atmosphere | Calm, nostalgic, introspective |
Why Structure Matters
Without structure, you rely on luck. With structure, you get consistent, reproducible results. Here’s what changes:
- Reproducibility — generate similar results every time by keeping fields locked
- Iteration — change one field (e.g., lighting) while keeping everything else identical
- Quality scoring — measure and improve prompt quality systematically
- Reusability — save prompts as templates and reuse them with variations
How to Write Each Field
1. Subject — Be Specific
The subject is the most important field. Avoid vague descriptions like “a person”. Instead, describe age, clothing, features, and distinguishing details.
✗ “A man”
✓ “A weathered fisherman in his 60s, wearing a thick wool sweater, salt-and-pepper beard, deep smile lines”
2. Action — Show Movement
Actions bring images to life. Even subtle actions create visual interest. Describe what the subject is doing and how.
• “Leaning against a brick wall, looking at the sky”
• “Running through rain puddles, laughing”
• “Carefully pouring coffee into a ceramic mug”
3. Environment — Set the Scene
The environment provides context and depth. Describe the location, time of day, weather, and background elements.
4. Style — Define the Look
Style controls the artistic direction. Use specific references: cinematic, oil painting, watercolor, cyberpunk, film noir, Studio Ghibli, etc.
Avoid: Phrases like “ultra realistic” or “best quality” can actually worsen results with modern models. Be specific about what kind of realism you want instead.
5. Lighting — Control the Atmosphere
Lighting is what separates amateur prompts from professional ones. Describe the light source, direction, quality, and color temperature.
• “Golden hour sunlight from the left, warm tones”
• “Harsh overhead fluorescent lighting, cold blue cast”
• “Soft diffused window light, natural shadows”
• “Dramatic Rembrandt lighting with a single key light”
6. Camera — Technical Precision
Camera settings tell the AI about perspective, focus, and framing. Use real photography terminology: focal length, aperture, lens type.
• “Canon EOS R5, 50mm f/1.4, shallow depth of field”
• “Wide-angle 16mm, low angle shot, deep focus”
• “Macro lens, extreme close-up, bokeh background”
• “Aerial drone shot, bird’s eye view, tilt-shift effect”
7. Emotion — Set the Mood
The emotion field is subtle but powerful. It influences color grading, composition, and overall feel. Use one to three mood words.
Complete Prompt Examples
Subject: A woman in her 30s wearing a vintage red dress, dark curly hair
Action: Sitting alone at a small round table, stirring coffee
Environment: A dimly lit Parisian café at night, rain on the windows
Style: Cinematic, film noir, moody
Lighting: Warm tungsten light from a table lamp, cool blue ambient
Camera: 85mm f/1.2, medium close-up, shallow depth of field
Emotion: Nostalgic, contemplative
Subject: A premium leather watch on a dark marble surface
Action: Placed at a slight angle, strap coiled naturally
Environment: Minimalist studio with seamless dark background
Style: Commercial product photography, editorial
Lighting: Soft key light from top-left, subtle rim light, controlled reflections
Camera: 100mm macro, f/5.6, focus stacked, crisp detail
Emotion: Luxurious, refined
Subject: An ancient stone castle on a floating island
Action: Waterfalls cascading off the edges into clouds below
Environment: Above a sea of clouds at sunset, distant mountains
Style: Digital matte painting, epic fantasy, concept art
Lighting: Golden hour from behind, volumetric god rays through clouds
Camera: Wide-angle 24mm, panoramic, deep focus
Emotion: Awe-inspiring, majestic
How to Improve Your Prompts
- Fill all 7 fields — empty fields leave the AI guessing. More detail = more control.
- Use specific terms — “85mm f/1.2” is better than “blurry background”.
- Avoid negative framing — say what you want, not what you don’t want.
- Iterate — change one field at a time. Lock the rest. Compare results.
- Use quality scoring — D.O.M.A. rates each prompt 0–100. Aim for 80+ (green).
- Study what works — browse the D.O.M.A. Gallery for inspiration.
Pro tip: Use D.O.M.A.’s built-in AI Chat assistant to help fill in fields you’re unsure about. It analyzes your current prompt and suggests improvements for each field.
Common Prompt Mistakes to Avoid
- “Ultra realistic, 8K, best quality” — these buzzwords add noise, not quality
- Vague subjects — “a person” gives random results; be specific
- No lighting description — lighting is what makes or breaks an image
- Missing camera info — without focal length, the AI picks randomly
- Too many conflicting styles — pick one clear direction