AI Prompt Engineering Guide
Comprehensive guide on writing high-quality AI image prompts. Learn about JSON vs natural language, style modifiers, and composition techniques.
1. Subject Specificity
The subject is the heart of your prompt. Instead of just saying "a cat," describe its breed, color, and action. For example: "A sleek Siamese cat wearing a holographic tuxedo jumping over a neon-lit puddle."
2. Style & Medium
Define the artistic medium to set the visual standard. Use terms like "Oil on canvas," "Cyberpunk digital art," "Ukiyo-e woodblock print," or "Cinematic 1950s photography."
3. Lighting & Mood
Lighting dictates the atmosphere. Modifiers like "Golden hour," "Volumetric lighting," "Dramatic Chiaroscuro," or "Vaporwave neon palette" can transform the same subject into vastly different moods.
4. Composition & Framing
Act as a director. Specify camera angles and lenses: "Macro shot," "Wide-angle aerial view," "Low angle hero shot," or "Bokeh depth of field."
5. JSON vs. Natural Language
Modern models like Midjourney and DALL-E 3 respond best to descriptive natural language sentences. However, technical models like Stable Diffusion or Flux often utilize JSON-like prompting or weighted tokens (e.g., [subject:0.8]) for precise control over parameters like seed and CFG scale.