Predicting video from a still image is a fraught task. Predicting interaction from a linear video is another fraught task.

It would be surprising if someone had the ability to do this.