Prepare your two references
Choose a character image with a clear face and visible body position. Pair it with a short video where the subject stays in frame and the movement is easy to see.
VISUAL INSPIRATIONAI motion control / Kling 3.0
AI motion control. Borrow the movement.
Turn a character image and a reference performance into a new video. Guide the motion, choose the character orientation, and decide whether to keep the original sound.
2 inputsCharacter + motion reference
3–30 secWith Match Video
MP4 / MOVReference video formats
ONE IMAGE. ONE PERFORMANCE.
The image sets the look. The clip guides the action.
Use motion control when you already have a gesture, pose change, or performance worth reusing. Start with a short, clearly framed reference so you can judge the transfer.
INPUT 01Your character
A clear image of the subject you want to animate.
INPUT 02Your movement
A video showing the performance to follow.
GENERATED OUTPUTA new performance
Review the character, motion, and surrounding scene.
Choose orientation and sound
Open Kling 3.0 Motion Control. Upload the image and video, choose Match Video or Match Image, and set Original Sound. Check the credit estimate before generating.
Review the complete clip
Check the face, hands, clothing, and background as the subject moves. If a section drifts, simplify the reference or trim to a cleaner movement before trying again.
SET UP THE SHOT
A few controls. A clearer starting point.
The reference video's length determines the motion-controlled clip. The workspace uses Pro mode and offers orientation and source-sound controls.
Match Video
Follow the reference video's character orientation. Use a reference from 3 to 30 seconds.
Match Image
Use the character image for orientation. Keep the reference between 3 and 10 seconds.
Original Sound
Keep the audio from the source clip, or switch it off to prepare for a separate soundtrack.
CREATIVE STARTING POINTS
Choose one movement to get right.
Start with a short performance that keeps the subject clearly visible. Build a more complex sequence once the first movement works.
A character performance
Pair a character portrait with a short head turn or expression. Start with a face angle close to the reference and inspect the eyes and mouth as they move.
A gesture or dance detail
Choose a full-body image and a clip where arms and legs stay visible. Try a simple gesture before fast footwork or a sequence with several turns.
A presentation moment
Use a clear reference of a person pointing, greeting, or reacting. Keep the original sound when it supports the performance, then check timing and visual consistency.
REFINE THE INPUTS
When the movement needs another take.
Motion transfer is generated, so every detail may not survive unchanged. Adjust one input at a time to understand what improves the result.
The face changes during a turn
Try a sharper character image with a similar starting angle. Use a reference with a smaller turn and fewer moments where the face is hidden.
Hands or limbs become unstable
Choose a clip with the whole gesture in frame. Avoid rapid cuts and overlapping limbs in the first test; a short, continuous movement is easier to assess.
The framing feels wrong
Check the orientation setting and the subject's scale in both inputs. A close-up portrait and a distant full-body performance can give the model conflicting information.
Clothing or scenery changes
Use a simpler background and inspect the entire output. This workflow generates a new video; it cannot guarantee preservation of all source details.
THE DETAILS
A few things
to know.
From your first idea to the final result.
What is AI motion control?
Motion control uses a reference video to guide the movement of a character in a separate image. In Unify, choose Kling 3.0 Motion Control, upload both inputs, and generate a new clip. The reference supplies the performance; the image supplies the character's appearance.
How long can my motion reference be?
Use a clip of at least 3 seconds. Match Video allows up to 30 seconds; Match Image allows up to 10 seconds. Longer uploads open the trimmer so you can choose a shorter section. Any account-specific trial limits are shown in the workspace.
Which video files can I upload?
Upload an MP4 or MOV file up to 100 MB. Both the width and height must be between 340 and 3850 pixels. Use a clear, continuous performance with the subject visible throughout the part you want to transfer.
What do Match Video and Match Image mean?
These controls choose whether the character orientation follows the reference video or the character image. Match Video supports clips up to 30 seconds; Match Image supports up to 10 seconds. Choose the orientation that fits the framing you want, then inspect the generated movement.
Can I keep the sound from the reference video?
Yes. Turn Original Sound on to keep the reference clip's sound, or turn it off when you want to add your own soundtrack later. This control retains source audio; it is not a dialogue or voice-generation control.
Is this the same as text-to-video or face swap?
Each workflow starts with different inputs. Motion Control follows an existing performance. Kling 3.0 text-to-video starts with written scene direction. The face swap app offers a guided photo identity edit and a video mode built on Motion Control. A generated video can change clothing, backgrounds, or other details as well as the face.
How much does a motion-controlled video cost?
The video workspace calculates the credit estimate from your uploaded clip length and the selected model's pricing. Review that estimate before generating. Start with a short section to check your references before committing to a longer clip.
YOUR NEXT IDEA STARTS HERE