Start with the image an AI video generator would need

Begin with a clean source image

A front-facing or three-quarter portrait with a clear subject gives the video model more stable information. Crowded scenes, hidden hands, and extreme poses leave more room for unwanted changes between frames.

Generate or choose the still first. Save a clean result you would want to animate when a real video provider becomes available.

Small motion often looks more convincing

Ask for a glance toward the camera, a slight smile, a change in posture, or gentle camera movement. Several complicated actions in a short clip can make the face and body drift.

The goal is a believable moment, not a feature film. A restrained prompt also makes it easier to preserve the character’s visual identity.

The same adult-content rules apply

Any future video generation will follow the same policy as images and chat. It will be limited to fictional adults and exclude minors, real-person sexualization, coercion, exploitation, and illegal content.