Skip to content

Motion Transfer · 12 engines · up to 1080p

AI motion transfer — animate an image with a video

Take the movement from one video and give it to the character in a still image. Kenerate's Motion Transfer tool reads the clip's body, hands and face, and renders your image performing it — on the engine that suits the motion.

  • Image + driving video
  • 12 engines
  • Up to 2 min
  • Up to 1080p

Updated by the

Made with Kenerate

Quick answer

Updated

What is AI motion transfer?

AI motion transfer takes the movement from one video — body, hands and often the face — and applies it to the character in a still image, so the image performs it. Kenerate AI's Motion Transfer tool does this on 12 engines, with driving clips up to 2 minutes and up to 1080p output.

Key facts

Inputs
1 image + 1 driving video
Video formats
MP4, MOV, WEBM, M4V, MKV
Upload size
About 100 MB
Engines
12 in 8 families
Highest output
1080p
Output
MP4 · no watermark

Real results

Motion worth transferring

A rooftop leap, a disco move, popping and locking, a tango demonstration and a surf pop-up — each a different kind of motion: athletic, rhythmic, precise, instructional, full-body.

Example clips from image to video, model named on each card. Motion transfer uses the movement of your driving video instead of a prompt.

How to

Transfer motion in five steps

Time needed
Steps
4 steps
  1. Step 1

    Choose the driving video

    The movement you want copied. MP4, MOV, WEBM, M4V or MKV, about 100 MB at most. Upload it, or pick from your History or library — links aren't accepted.

  2. Step 2

    Choose the image

    The character that will perform it. Clear JPG or PNG; on LTX-2 Control the image is optional.

  3. Step 3

    Pick the engine for the motion

    SteadyDancer for dance, Kling for realistic detail, LTX-2 Control for structure, DreamActor v2 for performances.

  4. Step 4

    Set the engine's options

    Resolution, prompt, seed, and where offered: keep sound, frame rate (Kenerate Motion), pose/depth/edges (LTX-2 Control).

Kenerate AI Motion Transfer with a full-body photo of an explorer character added as the character image

Open Motion Transfer and upload a clear photo of the character, or choose one from your history.

1 / 5 · Pick a character image

Watch the tour · 1:19

Text, image & reference to video, models, edit, extend & chat

Chapters

Engines

Which engine for which kind of motion

Engines read motion differently — as a skeleton, as structure, or straight from the pixels. Match the engine to what your clip is about.

  • SteadyDancer title card — a long
    Dance

    WaveSpeed

    SteadyDancer

    Input
    Image + video up to 2 min
    Output
    480p or 720p · silent

    Fast, athletic full-body motion that needs a stable, on-model character

    Open SteadyDancer
  • Kling 3.0 Pro Motion title card — a precise choreographed routine copied from a dancer onto a 3D character

    Kuaishou

    Kling 3.0 Pro Motion

    Input
    Image + 3–30 s video
    Output
    Follows the source · sound switch

    Realistic human detail — fingers, cloth, faces — on 3–30 s clips

    Open Kling 3.0 Pro Motion
  • LTX-2 Control title card — a pose skeleton
    Structure

    Lightricks

    LTX-2 Control

    Input
    Video up to 20 s (+ optional image)
    Output
    480p, 720p or 1080p · keep, new or mute sound

    Pick pose, depth or edges as the guide; the image is optional, up to 1080p

    Open LTX-2 Control
  • Wan 2.2 Fun Control title card — a control map guiding a video frame of a fictional adult dancer

    Alibaba

    Wan 2.2 Fun Control

    Input
    Image + video up to 2 min
    Output
    480p or 720p · silent

    Structure-guided transfer that restyles the clip from your image

    Open Wan 2.2 Fun Control
  • DreamActor v2 title card — a fictional adult dancer mid-move with an expressive face

    ByteDance

    DreamActor v2

    Input
    Image + video up to 30 s · no prompt
    Output
    Follows the source · keeps sound

    Performance transfer: body, face and lip movement, no prompt

    Open DreamActor v2
  • Kenerate Motion title card — a dancer's silhouette whose motion is being copied onto a stylised character

    Kenerate

    Kenerate Motion

    Input
    Image + video (first 20 s used)
    Output
    720p or 1080p · 24/48 fps · sound switch

    1080p output and a frame-rate choice: original, 24 or 48 fps

    Open Kenerate Motion

Outputs are MP4 without a watermark. SCAIL, Wan, Kling, Kenerate Motion and SteadyDancer trim an over-long clip; DreamActor v2, PixVerse Mimic and LTX-2 Control refuse it.

Tips

Clips that transfer well

The driving video does most of the work, so a few habits when filming it pay off.

  • Film on a tripod — camera shake reads as body movement.

  • Plain background, contrasting clothes: limbs are easier to track.

  • Keep the whole body in frame for full-body moves.

  • Avoid baggy sleeves that hide the elbows and wrists.

  • Trim dead time at the start and end — you pay per second of clip.

  • Transfer motion only onto people who agreed to it.

Example uses

How creators use AI motion Transfer

Illustrative examples of typical workflows, not customer reviews.

  • Physiotherapist demonstrating a shoulder exercise while a phone records

    Physiotherapist

    A physiotherapist can record an exercise once and transfer it onto a neutral illustrated figure for patient handouts, without showing a real patient.

  • Instructional designer reviewing a presenter video beside an illustrated host

    E-learning designer

    An e-learning designer can turn a filmed demonstration into the same demonstration by the course's illustrated host.

  • Football coach demonstrating a dribbling drill between cones

    Sports coach

    A coach can transfer a pro-quality movement drill onto a team mascot to make drill videos kids actually watch.

FAQ

AI motion transfer questions

Is motion transfer the same as motion control?

Mostly, yes. 'Motion control' is the name Kling uses for its version, and many tools now use the two terms interchangeably for making an image copy a video's movement. Kenerate's tool is called Motion Transfer and includes Kling 3.0's motion control alongside 10 other engines, so both searches lead to the same place.

Link to this answer

What files can I use for AI motion transfer?

The driving video can be MP4, MOV, WEBM, M4V or MKV, up to about 100 MB; Kling engines take MP4 or MOV. The character image should be a clear JPG or PNG — WEBP is best avoided. Upload them, or pick either from your History or library; external links are refused.

Link to this answer

What resolution does motion transfer output?

It depends on the engine. SCAIL, Wan and SteadyDancer offer 480p or 720p; PixVerse Mimic 360p, 540p or 720p; Kenerate Motion 720p or 1080p; LTX-2 Control 480p, 720p or 1080p; Kling and DreamActor follow your source clip. For a sharper master, use the result's Upscale to 4K hand-off.

Link to this answer

Can I change the frame rate?

On Kenerate Motion, yes: keep the original frame rate, or output 24 or 48 fps. The other engines don't offer a frame-rate setting. Kenerate Motion uses the first 20 seconds of the driving clip, so it suits short, smooth shots where you want 1080p and a specific frame rate.

Link to this answer

Can I transfer motion without a character image?

Yes, on LTX-2 Control. It re-renders the driving clip guided by its pose, depth or edges, and the image is optional — a prompt alone can restyle the whole scene. Every other engine needs one character image, because the image is what the movement is transferred onto.

Link to this answer

Do I need to describe the motion in a prompt?

No. The driving video already shows the motion, so prompts describe the look instead — clothes, lighting, style. Wan 2.2 Animate 2 adds a 'Describe the motion' field for moves the clip shows poorly, and DreamActor v2 and PixVerse Mimic take no prompt at all.

Link to this answer

Whose images can I transfer motion onto?

Only images you have the right to use: yourself, characters you designed, or people who gave permission. Don't animate celebrities or strangers. Motion transfer can make anyone appear to do anything, so tell viewers the clip is AI-made when you share it.

Link to this answer

Three ways engines read motion

Skeleton methods are precise for bodies, structure maps keep scenes and silhouettes, and pose-free methods carry over faces and subtle performance. That's why one tool benefits from several engines.

  • Skeleton pose — the clip is turned into joint positions frame by frame; SCAIL's research adds depth so the pose stays consistent in 3D.
  • Structure maps — pose, depth or edge maps of each frame steer the render (LTX-2 Control, Wan 2.2 Fun Control).
  • Pose-free — the model learns motion straight from the video's pixels; ByteDance's DreamActor-M2 paper describes this approach.

Explore

Every Kenerate video tool

Generators, tools, models and guides — all in one studio.

See more of Kenerate AI in your Google resultsAdd kenerateai.com as a preferred source — Google shows more from your preferred sources in Top Stories and AI answers.

Transfer motion onto your image

One image, one driving video, 12 engines. Paid per second of clip, no subscription.

Tried it? Tell us how it went