For the broader Kling lineup, explore Kling 2.6, KlingAI Avatar 2.0, Kling O1, Kling 2.5, and Kling AI.

Tip: keep the prompt focused on scene details (lighting, location, objects, atmosphere). Let the motion reference do the heavy lifting for performance.

Best practice: use a motion reference with moderate speed and minimal displacement, and avoid cuts for a steadier result.
Tip: keep the character’s hands visible in the image reference, and choose a motion reference with stable framing (no camera shake) so the model can track finger movement and contact points cleanly.
Avoid describing the main action in the prompt if it already exists in the motion reference (e.g., “dance, wave, point”). Use the prompt for scene intent instead: location, lighting, props, mood, and background activity. This reduces conflicting instructions and keeps motion cleaner.
Add a short identity lock line so the character stays consistent: “same face, same outfit, same hairstyle, consistent proportions.” If you need stable hands, make sure hands are visible in the image reference and avoid prompts that introduce gloves, extra accessories, or sudden wardrobe changes.
For best results, keep the subject framing aligned: full-body image + full-body motion reference, or half-body + half-body. Leave extra space around the subject for big moves (jumps, spins, wide arm gestures) to prevent awkward cropping or missing limbs.
Keep negative prompts short and practical: “no extra fingers, no warped hands, no face distortion, no camera shake, no sudden cuts, no flicker.” If the output looks unstable, simplify the scene prompt and switch to a steadier, uninterrupted motion reference.
Use Kling Motion Control to produce repeatable character performances—drive actions with a motion reference, steer scene details with prompts, and iterate faster on ad-ready variants.
Generate Now