AI Video Generation: How Motion Control Works

Motion control is the technology behind Larpable's AI video generation. It's what allows you to take a static image of your AI identity and a reference video showing motion, and produce a new video where your identity performs that motion.

The Two-Input System

Motion control uses two inputs:

1. 1An image — defines what the character looks like (appearance, face, clothing, background)
2. 2A reference video — defines how the character moves (motion pattern, timing, gestures)

The system analyzes the motion in the reference video — how the head turns, how the body moves, facial expressions — and applies that motion pattern to the character in the image. The result is a video that combines the appearance of the image with the motion of the reference.

Why Two Inputs Instead of One?

Some AI video tools try to generate video from a text prompt alone. The results are often unpredictable — you don't control exactly what the character does or how they look. Motion control gives you precise control over both appearance and motion by separating the two into independent inputs.

This separation is what makes motion control ideal for AI influencer content. Your identity's appearance stays consistent (from the image), while you can vary the motion freely (by choosing different reference videos).

What Makes a Good Reference Video

The reference video is the most important factor in the quality of your generated video. The best reference videos have:

• Smooth, natural motion — not jerky or robotic
• Short duration — 5 to 10 seconds is ideal
• Good lighting on the person's face
• A clear view of the person throughout the video
• Simple backgrounds — complex scenes with occlusions are harder to reproduce

What Makes a Good Image Input

The image should be:

• A clear, well-lit photo of your AI identity
• Framed similarly to the reference video — matching shot type (close-up, medium, etc.)
• High quality — a realistic generated photo works best
• In a starting pose that's somewhat similar to the reference video's first frame

Limitations to Understand

Motion control is powerful but has limitations:

• Very fast or complex motions may not reproduce accurately
• The video's background comes from the image, not the reference video
• Extremely long videos are harder to generate consistently
• Results depend on the quality of both inputs

Next Steps

Ready to try it? Read our guide to creating AI influencer videos and the video generation docs.