Clarity beats keyword accumulation
A long prompt is not automatically a precise prompt. Better results come from a controlled description in which every phrase has a role. The subject establishes what must appear, the environment supplies context, composition defines how the scene is viewed and lighting determines mood and volume. Style should refine this structure, not replace it.
Start with an unambiguous subject
Name the subject in concrete terms and include the action or state that matters. Replace vague descriptions with observable traits. For a product image, define the object, material and placement. For a portrait, define the adult subject, pose, wardrobe and expression. This prevents secondary style language from competing with the central idea.
- Use concrete nouns before adjectives
- Describe one primary action or state
- Include only identity-defining attributes
- Avoid contradictory roles or environments
Control composition before decoration
Composition phrases shape the image more fundamentally than decorative descriptors. State whether the image is a close-up, a wide architectural view, a centered packshot or an asymmetrical editorial frame. Add the camera angle, the distribution of negative space and the desired depth. These choices help the model understand where visual attention should go.
- Framing: close-up, medium shot or wide view
- Viewpoint: eye level, low angle or top-down
- Layout: centered, symmetrical or rule-of-thirds
- Depth: flat graphic space or layered realism
Finish with light, materials and rendering intent
Lighting explains the atmosphere and the physical form of the subject. Materials explain how surfaces should react to that light. Finish with the intended medium or rendering language, such as commercial product photography, editorial portraiture, cinematic architecture or polished interface visualization. Keep these instructions compatible with the scene you described.
- Direction and softness of the key light
- Surface qualities such as glass, metal, fabric or stone
- Color temperature and contrast level
- A single coherent rendering or photographic intent