Master the craft of describing images: subject, composition, lighting, color, and style.
Image prompting is the craft of turning words into pictures. Whether the tool is Midjourney, Flux, DALL-E, or Stable Diffusion, the same underlying skill applies: describe subject, composition, lighting, color, and style so precisely that the image you picture is the image you get.
Image prompting is a craft: describe subject, composition, lighting, color, and style so precisely that the image you picture is the image you get.
The subject is the anchor. Name it specifically: "a wooden sailboat" beats "a boat." Add the most important details that define it - material, color, age, state - without listing everything. Two or three defining details are enough.
Composition is about where things sit in the frame. Words like "close-up," "wide shot," "centered," "rule of thirds," "low angle," and "crowded street scene" direct the layout. If the image is framed wrong, fix the composition before touching anything else.
Lighting shapes mood more than any other single factor. "Golden hour," "harsh midday sun," "soft studio light," "candlelight," "neon at night" - each transforms the same scene completely. Name the light source and its quality.
Color sets emotion and style. "Muted earth tones," "vivid saturated colors," "monochrome blue," "warm sepia" all push the image in a different direction. Combine color words with lighting for a full atmosphere.
Style is the easiest way to change the whole feel: "watercolor," "photorealistic," "anime," "1980s poster art," "renaissance oil painting," "minimalist vector." The style word is often the single highest-impact token in the prompt.
A complete image prompt stacks these layers: "a wooden sailboat at golden hour, soft warm light, wide shot, muted earth tones, watercolor style, calm sea." Subject, lighting, composition, color, style - all five present.
The quality of the output scales with the quality of the words. Vague words produce generic images; specific, concrete words produce distinctive ones. Replace "beautiful" with a concrete detail and watch the difference.
A complete image prompt stacks these layers: "a wooden sailboat at golden hour, soft warm light, wide shot, muted earth tones, watercolor style, calm sea." Subject, lighting, composition, color, style - all five present.
Practice by describing one scene five ways: change the style each time, then change the lighting, then the composition. You will internalize which words move the image and which ones it ignores.
Overloading with conflicting details, neglecting lighting, fighting style words, and relying on negation instead of stating what you want are the common traps.
Describe one scene five ways - change style, then lighting, then composition - and you will learn which words move the image.