Image Prompt Engineer

The Problem

When a brief only says “high-end,” “magazine look,” or “the right mood,” image models can drift. The image-prompt-engineer skill treats prompts as executable visual specifications, breaking vague taste into subject, environment, lighting, composition, and technical parameters so outputs look closer to photographic stills rather than random collages.

How It Works

• Structured prompts: builds from subject, environment, lighting, style, and technical layers instead of relying on words like “nice light.”
• Photography translation: rewrites “blurry background” as “shallow depth of field, f/1.8 bokeh” and maps mood to golden-hour side light, Rembrandt lighting, or volumetric light.
• Platform adaptation: adjusts wording for Midjourney, DALL-E, Stable Diffusion, and Flux, including aspect-ratio flags, negative prompts, token weighting, or LoRA references.
• Workflow: confirms usage, target platform, references, and brand tone, then extracts lighting and composition cues, drafts layered prompts, and refines ambiguous or over-constrained phrases.

Boundaries

It is best for photographic, product, portrait, landscape, fashion, and commercial image prompts. It does not guarantee pixel-perfect consistency across models, nor replace real shooting, color grading, or brand-guideline decisions. When a platform ignores negative prompts or a model struggles with long descriptions, shorten the prompt and keep core fields such as subject, light, and composition.