How to build clearer image-to-image prompts

What an image-to-image prompt does

An image-to-image prompt combines written direction with one or more references. The written part explains what to preserve, what to change, and what kind of image you are trying to make. A useful prompt is not simply longer; it makes the relationship between instructions easier to understand.

Start with the intended result

Decide whether the goal is a character reference, a scene recreation, or a cinematic composition. That decision helps determine which details are essential and which should stay secondary. Write the result you need before collecting extra descriptive phrases.

Separate reference roles

Give each reference a job. One image may establish identity, another can support costume, and another can describe environment or pose. Keeping identity separate from costume and environment makes it easier to notice when two sources disagree.

Define camera and composition

Describe shot size, lens, camera height, angle, framing, and composition as their own direction. Camera choices affect how a subject appears, so they should not be buried inside identity or costume notes.

Avoid conflicting instructions

Check for incompatible viewpoints, body descriptions, lighting, or scenes. When priorities are clear, remove or rewrite details that compete with the intended result. An external image-generation service may still interpret the same prompt differently, so compare results by changing one controlled detail at a time.

How Scene Director supports this workflow

Scene Director organizes reference roles, scene controls, and Final Prompt or Compact Prompt output in one Windows workspace. It prepares English prompt text; you use that text and upload the chosen references in an external service.