Mastering the Writing of Prompts for Images
The difference between "okay" and "mind-blowing" often comes down to how you ask. In the world of AI, mastering the writing of promptsis key to success.
The true secret for mastering prompt is knowing what to write to make our idea come to life. Each task has its own "Golden Rule" to follow to get the best results and avoid common pitfalls.
1. The Builder: Creating New Images
Goal: Generating an image from scratch (Text-to-Image). The Golden Rule:
[Subject] + [Medium] + [Style] + [Details/Lighting]
Why this works: When creating from scratch, the AI starts with a blank canvas (random noise). It has zero context. It needs a complete blueprint to know what to build, how it looks, and where the light comes from.
- ❌ Bad Prompt: "A flying car in a city." (The AI has to guess everything else: Color? Year? Location? Style?)
- ✅ Good Prompt: "A sleek, silver flying car speeding through a city, digital concept art, cyberpunk aesthetic, neon lights reflecting on wet pavement."
2. The Director: Editing Images
Goal: Changing a specific part of an existing image (Inpainting/Editing). The Golden Rule:
[Action/Change] + [Specific Detail]
Why this works: The AI already sees the image. It knows there is a man standing in a city. If you describe the whole scene again ("A man in a city wearing a red shirt"), the AI might get confused and try to squeeze a tiny new man-in-a-city into the small shirt area you selected.
To fix this, focus only on the change.
- ❌ Bad Prompt: "A man wearing a red shirt standing in a city." (Confuses the AI with redundant info)
- ✅ Good Prompt: "Change the shirt to a bright red cotton t-shirt." (Clear, direct instruction)
3. The Stylist: Combining Multiple Images
Goal: Merging elements from different images into a new cohesive whole. The Golden Rule:
[Subject from Image A] + [Background from Image B] + [Unifying Context/Style]
Why this works: When combining images, the AI needs to know what to keep from each and how to glue them together. If you just upload two random photos without instruction, the AI might layer them awkwardly or produce a chaotic mess.
You need to be the "bridge" that explains the relationship.
- ❌ Bad Prompt: "Person and background." (The AI doesn't know how they fit together. Is the person floating? Standing?)
- ✅ Good Prompt: "A cinematic shot of [Person Description] standing in the center of a [Sci-Fi Background Description], ambient blue lighting from the environment reflecting on the subject, seamless integration."
Key Elements of a Great Prompt
No matter the task, these pillars always apply:
1. Be Specific, Not Vague
Why: "Vague" words leave too much room for the AI's random interpretations.
- ❌ Vague: "A nice dog."
- ✅ Specific: "A happy golden retriever puppy running through a field of sunflowers."
2. Describe the Lighting
Why: Lighting is the fastest way to set the mood and make the vibe of your image pop.
- Golden Hour: Warm, soft sunlight (great for portraits).
- Volumetric Lighting: God rays, hazy beams of light.
- Cinematic: Dramatic contrast.
- Neon: Bright, colorful, artificial light.
3. Mention Art Styles
Why: It constrains the AI to a specific visual vocabulary.
- Styles: Impressionism, Surrealism, Cyberpunk, Ukiyo-e.
The Prompt Keyword Cheat Sheet
Stuck on what to say? Mix and match these keywords to spice up your prompts:
💡 Lighting & Mood
| Keyword | Effect |
|---|---|
| Cinematic | Dramatic, movie-like contrast and depth. |
| Golden Hour | Warm, soft sunlight (best for outdoor portraits). |
| Bioluminescent | Glowing, alien-like light (great for sci-fi/fantasy). |
| Studio Lighting | Perfect, controlled lighting (best for products/headshots). |
| Volumetric Lighting | Visible beams of light ("God rays") cutting through dust/fog. |
| Neon | Bright colors, artificial light, for a futuristic and vibrant look. |
📷 Camera & Angles
| Keyword | Effect |
|---|---|
| Wide Angle | Captures more of the scene (good for landscapes/interiors). |
| Macro | Extreme close-up (good for insects, eyes, textures). |
| Bokeh / Depth of Field | Blurs the background to focus on the subject. |
| Drone View | High-up aerial perspective. |
| Low Angle | Looking up at the subject (makes them look powerful/large). |
| Fish-eye | Distorted, spherical view (good for action/skateboarding vibes). |
🎨 Art Styles
| Keyword | Effect |
|---|---|
| Studio Ghibli | Anime style, very detailed and realistic. |
| Cyberpunk | Futuristic, neon-lit, high-contrast, digital art. |
| Oil Painting | Visible brushstrokes, textured, classic art look. |
| Watercolor | Soft, bleeding colors, artistic and dreamy. |
| Line Art | Black and white, clean lines (good for coloring books/logos). |
| Pixel Art | Retro video game style (specify "16-bit" or "8-bit"). |
| Claymation | Looks like a stop-motion clay figure. |
🧱 Textures & Materials
| Keyword | Effect |
|---|---|
| Translucent | Semi-transparent, light passes through (glass, jelly). |
| Iridescent | Changing colors like a soap bubble or beetle shell. |
| Matte | Flat, non-reflective surface. |
| Rusty / Grungy | Old, worn, dirty texture. |
| Knitted / Woolen | Soft fabric texture. |
Next up: Learn how to create engaging short videos from simple text prompts using our Video Generator in our From Text to Motion: Video Generator Basics guide.