What is reference-driven prompting?

Definition

Reference-driven prompting begins with examples that communicate desired visual qualities more precisely than abstract wording. A workflow can analyze the reference and convert lighting, framing, composition, camera, color, and mood into structured prompt elements.

The goal is to learn from relevant attributes rather than copy the original work. Creators should preserve provenance, respect rights, identify which properties they intend to borrow, and combine references with their own shot requirements.

ELI5

Reference-driven prompting uses example images or clips to help explain what an AI generation should look or feel like. The reference can communicate lighting, framing, color, or composition that is difficult to describe from memory.

For example, a creator can find a night interior with useful lighting and turn its broad lighting ideas into a prompt for a different subject and scene. The workflow should not simply reproduce the original image.

Frequently asked questions

What can a visual reference contribute to a prompt?

It can contribute ideas about framing, composition, lighting, color, lens, camera angle, depth, texture, mood, motion, and pacing.

How is reference use different from copying?

Reference use extracts selected qualities for a new goal, while copying attempts to reproduce the original expression or protected details too closely.

Videos explaining reference-driven prompting

  1. Tim Simmons beside the headline Mystery AI Video Models