Does OutfitGen Work With Multiple People?

Why group photos are more difficult

In a single-person portrait, the model can usually connect one face, body, pose, and outfit. In a group, people may overlap, wear similar colors, stand at different distances, or be partly hidden. A request such as 'change the shirt to blue' may not say whose shirt should change.

Every additional subject adds boundaries the model must infer. Hands can cross another person's clothing, hair can cover a collar, and one person's arm can sit in front of another person's torso. Those overlaps make it easier for an edit to spill into the wrong area.

A dependable one-person workflow

• Duplicate the original so you always have an untouched master file.
• Crop around the person you want to edit while retaining enough of their body and surroundings to understand the pose.
• Describe the person by position and a visible feature, such as 'the person on the left wearing a black shirt.'
• Request one outfit change and explicitly ask to keep the face, pose, other people, and background unchanged.
• Repeat from the best available version for each additional person, checking the full image after every generation.

When to keep the full group in frame

Keep the full composition when spacing and interaction between people are essential. Use precise positional language and avoid asking for several different wardrobe changes in the same prompt. A generation may still alter more than intended, so plan for retries.

For client work or an important event photo, edit each person separately and composite reviewed results in a conventional editor. That adds a manual step, but it gives you more control over expressions, hands, garment edges, and the people who were not meant to change.