Can I use AI to make a model look at the camera if they are looking away?

Quick Answer

Yes, AI can often redirect a model's gaze when the face is clear and already close to front-facing. A gaze correction is still a generative edit that can redraw the eyes or nearby facial details; if the head is turned far away, the tool may need to invent hidden facial structure and can change the person's identity. Keep the instruction narrow, use an approved front-facing reference when available, and inspect the full image before publishing.

When is gaze correction a small edit rather than a new face?

A gaze correction is relatively contained when both eyes are visible, the head angle stays fixed, and only the pupils and iris direction need to move. The risk rises when one eye is hidden, the face is in profile, hair covers facial landmarks, or the request also changes head direction or expression. In those cases, the editor must construct details that the source photo never recorded.

This limitation applies broadly to generative image editing. OpenAI's current image-editor guidance says edits may extend beyond a selected area, while Adobe describes its Smart Portrait controls as generating new elements for gaze and head direction. A written request can define the intended scope, but it does not make the edit pixel-locked.

How should I correct a model's gaze in Nightjar?

In Nightjar, open the Edit tab, add the original photo to the board, and reference it as @image1. Start with one change:

On @image1, change only the eye direction so both eyes look toward the camera. Keep the head angle, expression, face shape, iris color, eyelids, catchlights, skin texture, hair, garment, product, lighting, crop, and background unchanged.

If you have an approved photo of the same person looking forward, add it as @image2 and say to use it only as the identity and eye reference for @image1. A clear source reference gives the editor evidence for features that may be obscured in the original instead of asking it to guess.

What should I review after an AI gaze edit?

Nightjar's built-in visual review can retry obvious eligible failures, including a clear identity substitution or catastrophic image defect, without consuming an extra Credit, Nightjar's unit for paid actions. It is not fine-grained eye, anatomy, or likeness approval, so a person should still compare every candidate with the source.

When should I regenerate, retouch, or reshoot the photo?

Regenerate a fresh shot when the head must rotate substantially, the first narrow edit changes identity, or several facial regions need repair. In Nightjar's Create tab, start a new Product Photography shot, select a reusable AI person called a Fashion Model to anchor identity, choose a Pose for the body, head, and gaze direction, then set a compatible Camera Distance for the crop. This gives the new output explicit references rather than forcing an eye-only edit to reconstruct a hidden face.

Use manual retouching when the rest of the photo is approved and an authorized forward-facing frame provides eyes that can be composited cleanly. Reshoot when exact real-person identity, anatomy, garment fit, product interaction, or named-talent approval is essential and no suitable source exists.

Before uploading or editing a real person's photo, confirm that you have rights to the source photograph and permission for the intended AI likeness use. Nightjar's Terms prohibit Input or Output that infringes third-party privacy or publicity rights; access to an editor does not supply those rights. Check the talent agreement and applicable local law for commercial use, reuse, and approval requirements.