← Back to image discovery

VERIFIED / Portrait & Character

Nine-Emotion Portrait Grid

A controlled identity-consistency test where only subtle facial emotion changes across nine frames.

PROMPT WORKSHOP

How to build the Nine-Emotion Portrait Grid prompt

Do not begin with one dense block. Confirm the visual goal, then add subject, composition, light, and constraints one layer at a time.

Start with one sentence

Base prompt

Nine-Emotion Portrait Grid. A controlled identity-consistency test where only subtle facial emotion changes across nine frames. 1:1 composition.
01

Iteration

Define the result

What changes: Theme, outcome, and aspect ratio only.

Why: Check whether the model understands the core image.

Nine-Emotion Portrait Grid. A controlled identity-consistency test where only subtle facial emotion changes across nine frames. 1:1 composition.
02

Iteration

Lock the subject

What changes: Add only subject and action constraints.

Why: A precise subject reduces identity and object drift.

Nine-Emotion Portrait Grid. A controlled identity-consistency test where only subtle facial emotion changes across nine frames. 1:1 composition. Define the subject, action, clothing, and fixed identity details precisely.
03

Iteration

Control composition and light

What changes: Add camera, composition, and lighting constraints.

Why: Composition controls hierarchy; light controls material and mood.

Nine-Emotion Portrait Grid. A controlled identity-consistency test where only subtle facial emotion changes across nine frames. 1:1 composition. Specify camera position, framing, visual hierarchy, light direction, and color temperature.
04

Iteration

Add texture and constraints

What changes: Add production detail and exclude: minor or childlike appearance, celebrity likeness, plastic skin, over-retouched face, extra fingers, malformed hands, duplicate limbs, unreadable text, logo, watermark, inconsistent identity, duplicate emotion, mismatched clothing, different camera angle

Why: The final pass addresses common model failures.

A precise three-by-three portrait contact sheet of the same fictional adult person, photographed from shoulders up against one neutral studio background. Identical hairstyle, clothing, lens, camera distance and soft clamshell lighting in every cell. Nine clearly different but natural emotions: calm, curious, amused, relieved, skeptical, focused, surprised, concerned and quietly joyful. Maintain exact facial identity and realistic skin texture across all frames. Thin even gutters, editorial casting-board presentation, no captions or text.

WHY IT WORKS

Prompt anatomy

Subject

Subject, action, clothing, and fixed identity details.

Reduces identity, pose, or object drift.
Composition

A 1:1 frame with explicit camera position and hierarchy.

Defines aspect ratio, camera position, framing, and hierarchy.
Lighting

Direction, softness, color temperature, and practical light sources.

Unifies material, depth, and mood.

TROUBLESHOOTING

How to fix weak results

The image looks like a generic AI template

Remove conflicting style terms and add one concrete camera behavior, material imperfection, and environmental detail.

Every iteration becomes a different image

Change one variable at a time and preserve the previous prompt and result.

The subject is correct but framing drifts

Move the camera and framing instruction immediately after the subject and remove conflicting lens terms.

FAQ

Common questions about this prompt

How do I start a Nine-Emotion Portrait Grid prompt?

Start with “Nine-Emotion Portrait Grid. A controlled identity-consistency test where only subtle facial emotion changes across nine frames. 1:1 composition.” and add subject, composition, light, and material constraints in separate passes.

Which image model should I use?

Use a current image model that handles detailed composition reliably. Compare prompt structure with default parameters before tuning settings.

Why does my result look different?

Image models are stochastic. Lock the subject and composition first, then change one variable at a time.

Can I replace the subject or location?

Yes. Replace one variable while keeping camera and lighting fixed, then evaluate before changing another variable.