Mise-en-scène

Mise-en-scène is about deciding what goes where within the frame when the camera captures a scene. Where the actors stand, where the light comes from, what appears in the background, what mood the colors and textures create—this French film term refers to the director's work of choosing and placing all of these elements. Simply put, it's like a director's hand arranging objects and people on a stage.

The same concept applies to AI-generated video. In the past, the main challenge was simply removing basic flaws—faces glitching, clothing changing mid-shot. Now that those flaws have decreased, the composition within the frame—where a figure stands, how the light falls—has emerged as the factor that shapes an impression of the work. In other words, mise-en-scène has become less a question of whether AI can generate a scene, and more a measure of how deliberately that scene has been designed.

When writing a video generation prompt, don't just describe the character's action—specify the camera position, the direction and color of the lighting, and even the props placed in the background. For example: 'A dark room, blue light from a window illuminating only the left side of the face, the figure standing at the right third of the frame.' Designing the composition in words like this beforehand makes the result feel far more like a deliberately crafted scene.