Skip to content
Jovoscope
Back to blog

AI Image Reference Prompts: Subject, Scene and Style

Nov 17, 2026Jovoscope

A practical way to separate image-reference roles and avoid conflicting instructions in a multi-image edit.

AI Image Reference Prompts: Subject, Scene and Style

The most useful image-reference prompt says what to take from each picture and what to leave behind. A portrait can supply a person’s appearance without supplying its background; a beach photo can supply a setting without supplying its people.

Give every image a job

Before uploading, finish this sentence for each image: “Use this for…” Useful answers include the main scene, the person, the clothing or the pose. If two images have the same job, decide which one should take priority.

A style reference needs especially clear wording. “Use the warm evening color from this image” asks for something different from “recreate this image.” Name the visible feature you want rather than relying on “same vibe.”

A brief for a travel portrait

Try this structure in AI Image Combiner:

Use the coastal path as the setting. Place the person from the portrait on the path, slightly left of center. Keep the person’s face, short hair and green jacket. Use the beach image for warm evening color only. Keep the sea visible on the right.

Adapt the description to your files. The instruction separates identity, placement and color, so you can tell which part needs changing after you see the output.

What to preserve belongs in the prompt

“Keep everything else” sounds precise but leaves the important details unnamed. If a necklace, jacket print or landmark matters, mention it. Keep the list focused: the model still needs a clear main task.

For a pose edit, begin with the original photo and describe the desired body position. A separate style image may add little. The pose reference guide explains how to choose a useful body reference.

Resolve contradictory instructions

A request for a close-up and full-length body cannot have equal priority in one frame. Neither can “keep the original room” and “put the person on a beach.” Choose the final composition first, then remove instructions that belong to another version.

If you want two different directions, write two briefs and keep the source images unchanged. Compare the outputs against the brief, not just against each other.

Describe placement using the scene

“On the second step,” “to the left of the doorway” and “behind the table” are easier to judge than “somewhere natural.” Include whether the whole body should be visible. For a new person in a group, describe who they stand next to and whether they are in the front or back row; see adding a missing person.

A final prompt check

Read only the first sentence. Does it identify the intended edit? Then read the rest and remove duplicate adjectives. One clear action, named references and a few preservation details make a useful starting brief.

Use AI Image Combiner for a shared scene. For task selection before you upload, read the travel photo editing checklist.