A prompt is the text instruction or description a creator provides to an AI generation model to direct what it produces. It is the primary interface through which human creative intent is communicated to the model, translating a visual idea, scene description, or stylistic direction into language that the model can interpret and act on. The quality, specificity, and structure of a prompt have a direct and significant impact on the quality and relevance of the generated output.

Prompts for image and video generation typically describe the subject, setting, mood, lighting, camera angle, style, and any other visual elements relevant to the intended output. A minimal prompt might be a few words - "a foggy forest at dawn" - while a detailed prompt might specify camera position, lighting quality, color palette, time period, artistic style, and specific compositional elements across several sentences. Models respond to both the semantic content of prompts (what concepts are described) and the specific vocabulary used (certain terms have strong associations in the model's training data that influence output reliably). Different models have different prompt sensitivities and respond best to different levels of detail and different vocabularies.

Writing effective prompts is a skill that develops through practice and experimentation. Understanding how a specific model interprets language, which descriptive terms produce predictable results, and how to structure descriptions to prioritize the most important elements all improve with experience. On Morphic, prompts drive generation across all supported models, making prompt craft one of the most directly impactful skills for getting high-quality, consistent results.