See the Motion Before You Generate
Real motion control example
Character image + motion reference
Generated result
City character performance
Character image
Motion reference
Image + motion reference + generated video
Open any example to compare the character image, motion reference, model, prompt, and generated result in one place.
A full-body animated character follows a confident pose and gesture performance in a vertical city scene.
Kling Video 3.0 · Motion Control
Generation prompt
No additional scene prompt. The character image and reference performance provide the motion direction.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
Two dressed cats copy the same dance timing while the camera, animal paws, fur, clothing, and scale remain consistent.
Kling Video 3.0 · Motion Control
Generation prompt
Two cats are dancing while keeping the original appearance consistent. Follow body movement, facial expression, mouth movement, eye expression, and timing from the reference. Keep the camera stationary, preserve realistic paws, fur, clothing, body proportions, natural blinking, and sunny park lighting.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
A photorealistic cat in a green alien costume follows a full-body dance reference while paws, fur, clothing, scale, and the fixed fantasy scene stay consistent.
Kling Video 3.0 · Motion Control
Generation prompt
The cute cat dances in sync with the reference. Follow the body movement and facial expression, keep the camera stationary, preserve animal paws, fur, costume, scale, and position, and use natural fabric motion with cinematic warm lighting.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
A sporty character transfers a guarded stance, block, straight punch, step, and finishing pose from one action reference.
Kling Video 3.0 · Motion Control
Generation prompt
A red-haired character performs a short martial arts sequence with smooth, decisive motion: stable pose, slight step, guard stance, block, straight punch, and strong finish. Keep the face stable, anatomy correct, hands and legs clear, and the modern background consistent.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
A miniature child standing on a palm receives subtle breathing, blinking, smiling, hair movement, and a slow hand wave.
Kling Video 2.6 · Motion Control
Generation prompt
A tiny pocket-sized child stands on a human palm, viewed from above. The child gently blinks, breathes naturally, smiles softly, and slowly waves one hand. Hair moves slightly in a light breeze while the hand remains steady. Photorealistic cinematic lighting.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
An illustrated sword character follows a fast full-body reference while keeping the anime design and dramatic environment.
Kling Video 2.6 · Motion Control
Generation prompt
No additional scene prompt. The anime character image defines the visual identity and the motion video supplies the sword performance.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
A stylized mech character follows a longer full-body action reference through turns, steps, arm movement, and combat poses in a ruined city.
Kling Video 3.0 · Motion Control
Generation prompt
No additional scene prompt. The mech character image defines the robot design and ruined-city look while the motion video supplies the full-body action and battle timing.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
Two stylized characters follow synchronized dance timing while the fixed camera, character scale, and colorful stage remain stable.
Kling Video 3.0 · Motion Control
Generation prompt
Two stylized characters dance in sync. Follow the body movements in the reference video, keep the camera stationary, preserve both characters’ scale and position, and avoid zooming or distortion.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
A vertical creator-style dance clip focuses on face consistency, realistic skin, grounded feet, and natural indoor lighting.
Kling Video 2.6 · Motion Control
Generation prompt
Preserve the source face, body shape, outfit, skin texture, and natural proportions while following the reference dance. Keep the subject grounded, match the room lighting and shadows, avoid edge artifacts and body warping, and retain an authentic creator-video look.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
Four distinct performers follow one synchronized group-dance reference while spacing, identities, body count, stage lighting, and camera framing remain controlled.
Kling Video 3.0 · Motion Control
Generation prompt
Keep the same four distinct dancers and consistent faces throughout the routine. Follow the reference movement with clear space between performers, no merging or disappearing bodies, stable group composition, full-body movement, and naturally changing stage lights.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
A controlled martial-arts example for slower weight shifts, open-hand gestures, torso rotation, and balanced poses.
Kling Video 3.0 · Motion Control
Generation prompt
No additional scene prompt. Use the character image for identity and the tai chi reference for slow, continuous body direction.
This example uses one character image and one motion reference video. Open the media comparison to review both inputs beside the generated result.
Each case pairs one character image with one reference video. Open a card to inspect the source image, motion clip, model, prompt, and final video together. The reference supplies visible action and timing, while the image supplies character identity, proportions, clothing, and visual style.
The collection covers photo-to-dance video, pet dance animation, anime motion transfer, martial-arts motion capture, subtle character gestures, and creator-style social video. Results still depend on source clarity, framing, occlusion, motion speed, body-structure differences, model choice, orientation, and background mode.
Use the branch that matches the final video you want to create.
Compare single-dancer references, full-body framing, and short social dance outputs.
Use close, front-facing references for visible mouth, expression, and presenter timing.
Transfer gestures, posture, running, martial arts, and other single-character actions.
Start with the simplest possible pair, then increase motion complexity after the identity remains stable.
Decide whether the output is primarily dance, talking performance, or general action transfer.
Use similar portrait or landscape framing and keep the main performer visible.
Review a short clip before increasing the duration or using a more complex reference.
The same motion-control engine supports multiple user vocabularies and content formats.
Make a photo or humanoid character perform a reference dance.
Make a character appear to deliver a short talking performance.
Record your own reference and direct the character with your body and face.
Inputs, recommended setup, and the most common fixes—summarized in one compact guide.
Start with one clear character image and the cleanest section of your reference video.
