Lesson Content

Structured prompt layers.

Advanced prompts have a clear structure: (1) Subject – what is the primary visual subject; (2) Action/Motion – what is the subject doing or how is the scene moving; (3) Environment – what is the setting and background; (4) Style – art style and aesthetic category; (5) Mood/Atmosphere – emotional quality and lighting; (6) Quality modifiers – resolution, detail level, cinematic quality; (7) Camera – shot type, angle, movement. Building each layer deliberately produces more controlled results than freeform prompt construction.

Precision descriptor selection.

Precision descriptors have a narrow semantic range – they can be interpreted only one way. "Cinematic" is imprecise: it can mean dramatic, high-production-value, a specific look, or a color grade. "Anamorphic lens flare, high contrast cinematic color grade" is precise: it describes a specific visual quality. Audit prompts for vague descriptors and replace them with precise alternatives when generation results are inconsistent with intent.

Systematic refinement strategy.

When iterating, change one variable per iteration where possible – to understand what change produced what result. Changing three descriptors simultaneously makes it impossible to know which change improved or degraded the output. Single-variable iteration is more labor-intensive but produces more reliable, transferable learning about what prompt elements control what visual qualities.

Prompt architecture documentation.

Document the architecture of successful advanced prompts – the specific layers, the precision descriptors that were most effective, the iteration path that led to the final result. Advanced production is built on this institutional knowledge; undocumented prompt architecture is knowledge that doesn't transfer across sessions or team members.