Have a stunning Midjourney still, portrait photograph, or digital artwork that you want to animate? Learn how to direct camera movement, temporal physics, and subject animation for Kling AI, Runway Gen-3, and Luma Dream Machine.
In Image-to-Video (I2V) mode, the AI model already knows what the scene looks like because your first frame is locked. Therefore, your prompt should not re-describe the image (e.g. do not write "a man with brown hair wearing a jacket"). Instead, focus 100% on kinetic motion: camera trajectory, subject choreography, wind/water physics, and lighting changes.
Direct how the virtual lens navigates the 3D space: Dolly In, Pedestal Up, Tracking Left, or 360 Orbital sweep.
Describe physical action: walking forward, smiling, reaching out a hand, or speaking dialogue naturally.
Animate ambient elements: billowing fog, drifting embers, rustling leaves, falling raindrops, or splashing ocean waves.
Instruct dynamic lighting: a passing car's headlights washing across the scene, sunset fading to dusk, or flickering neon.