Turn character photos and a motion reference into a video. Choose a community or classic template, or upload your own clip, then generate with the matching model.
Creations from this workspace appear here.
Find your next video idea. Choose a ready-made trend and make it your own.
Motion Transfer AI animates a character photo using the movement in a reference video. The photo supplies the appearance; the video guides gestures, expressions, and body movement. Pair a recorded performance or template with a photo that shows the body parts involved. You get a new character video to preview and download, with movement guided by something you can show instead of timing you need to describe.
Prepare two references, choose the settings that fit your clip, and review the result in the same workspace.
Step 1
Choose a motion template or upload one continuous video, then add your character photo. Match the crop to the action: include hands and elbows for a gesture, or the whole body for a dance. Preview the motion before selecting it.
Step 2
Select a model, open Settings, and choose a supported resolution. For Kling, also select Match photo or Match video. Check the clip length and credit estimate before generating; each of these choices can affect the job.
Step 3
Sign in when prompted, then generate with both inputs ready. Find the result in My creations, play it through, and check the face, hands, and pose transitions. Download the clip you want to keep; available results can also be reopened from account history.
Use Motion Transfer AI when a specific performance matters: the order of a gesture, the movement of a character, or the framing of a short clip.
01
A wave includes direction, timing, and the moment the arm returns to rest. Those details can be awkward to explain in text. With Motion Transfer AI, a reference clip shows the sequence you want. Choose one complete action with a clear beginning and end. A short, readable performance gives you a more practical starting point for reviewing the result.
02
A longer recording may contain one gesture you want to reuse. Trim it to that continuous action before uploading to Motion Transfer AI. Leave enough footage to show the start and finish, and remove unrelated sections. You can supply a prepared clip without asking the model to interpret several different scenes.
03
Model choice, resolution, and clip length all affect the credits needed. The workspace shows an estimate before you generate, so you can adjust the setup to fit your balance. Start with a short Motion Transfer AI test when you want to check an unfamiliar pair of references.
Explore three ways to use a photo and a performance together. The example clips show the kind of movement to look for when choosing your own references.
Car performance
01
Make a character wave, give a greeting, or follow an expressive arm movement for a short social post. Choose a reference in which the hands stay visible throughout the gesture, then pair it with an upper-body photo that includes the elbows. Motion Transfer AI gives the action a visual reference. Before sharing, watch the fingers and wrists as they cross the face or torso; overlapping limbs are a useful place to spot problems.
02
Try a realistic portrait or an illustrated character with the same motion reference. This is useful for exploring a character introduction or comparing visual styles without changing the intended gesture. Use a similar crop for each photo so your comparison stays focused on appearance. Motion Transfer AI creates a new interpretation each time, so inspect the face and outfit as well as the movement. Keep the version that best fits your project.
03
A smile, head turn, or small upper-body movement can give a still portrait a short performance. Start with a clear face and a motion clip whose head angle stays readable. A close photo fits a close movement; it leaves little information for a full-body dance. For a Motion Transfer AI portrait, review the eyes, mouth, and outline of the face through the whole clip, especially when the head turns away from the camera.
Use these paired examples to check framing and subject visibility before uploading. They illustrate how to choose your source files.
For full-body motion, include the head, hands, hips, and feet, with space around the limbs. For a portrait gesture, use a closer crop that still contains every moving body part. A tight headshot gives Motion Transfer AI little visual reference for legs or outstretched arms.
Recommended
Avoid
Choose one clearly visible person and a continuous shot. Watch for hands leaving the frame, people crossing in front, and objects covering the body. Motion Transfer AI needs a readable reference throughout the action, including the moments between the poses you like.
Recommended
Avoid
Play the whole motion video before using it. Trim away unrelated footage and scene changes, keep a steady camera, and choose a pace at which you can still follow the hands and feet. If a section is heavily blurred or hidden, try a clearer take.
One credit balance for images, videos and motion transfer.
For your first ideas.
$100.0 charged yearly
2,400 credits/year
Up to 400 images
Up to 133 videos
For a steady creative flow.
$360.0 charged yearly
21,600 credits/year
Up to 3,600 images
Up to 1,200 videos
For more creations, more often.
$599.0 charged yearly
48,000 credits/year
Up to 8,000 images
Up to 2,666 videos
Try before you subscribe
Prices in USD; taxes and total shown at checkout. Subscriptions renew automatically; cancel in Billing settings. Subscription credits expire each period; one-time credits never expire.
Refunds: request within 72 hours. Used purchased credits and applicable non-refundable processing costs (up to 6%) are deducted. Terms · Refund Policy
Practical answers about files, model settings, credits, and reviewing your finished video.
For Kling 2.6 and Kling 3.0, upload one JPEG or PNG image up to 10 MB. Wan 2.2 also accepts WebP. Seedance 2.0 and Seedance 2.5 accept multiple reference images plus a motion video. Check the formats and upload limits shown for your selected model. Seedance reference videos can be 2–15 seconds for 2.0 or 2–30 seconds for 2.5; Kling and Wan accept 3–30 seconds, with a shorter limit for some Kling settings. A supported file format does not compensate for a blurry face or hidden limbs.
Kling 2.6 and Kling 3.0 accept 3–10 seconds with Match photo, or 3–30 seconds with Match video. Wan accepts 3–30 seconds. Trim longer recordings to one continuous action before adding them to Motion Transfer AI. The estimate follows the actual clip duration, so a shorter reference can also help you test an idea with fewer credits.
These Kling settings choose whether character orientation follows the photo or the video. Match photo supports reference clips up to 10 seconds; Match video supports up to 30 seconds. Choose the setting that fits the reference and the composition you want, then inspect the result. Motion Transfer AI does not make every background detail or camera movement match exactly.
The workspace offers Kling 2.6 Motion Control, Kling 3.0 Motion Control, Wan 2.2 Animate Move, Wan 2.2 Animate Replace, Seedance 2.0, and Seedance 2.5. Kling and Wan keep one-photo motion transfer. Seedance is for multi-image plus video templates. Start with the default Kling model if you are exploring, then switch when a template or reference set needs Seedance.
The motion comes from the reference video. Kling 2.6 and Kling 3.0 offer optional Scene details for a scene note; Wan Move and Wan Replace have no prompt field. Choose the motion first instead of relying on text to invent a different gesture. If you add a scene note, keep it consistent with the photo and the action you are supplying.
Yes. Community templates load a motion clip, numbered reference images, and a prompt without generating. Classic templates still load only the motion video and keep your photo. Preview cards show community examples, not this site’s generations. You can replace each image and keep the @ImageN order.
New accounts receive 18 starting credits. That covers a Kling 2.6 720p job with a clip of exactly 3 seconds; a longer clip or another setting can require more. Weekly check-ins offer up to 60 additional credits across the reward cycle. These rewards are separate from the starting grant. Check the estimate and your balance before generating.
Use your own likeness or someone you are explicitly authorized to depict, and make sure you have permission to use the source media. The same requirement applies when making a character greeting, a dance, or a clip for a client. Using Motion Transfer AI does not grant rights to a person, image, performance, or character. Review the Use Policy before submitting.
Motion Transfer AI generates a new video from your references. Facial details, clothing edges, lighting, and background objects can shift as the character moves. Start with a sharp, well-lit photo and a simpler motion clip. If an important detail disappears when the subject turns, try a reference with a more readable angle. Treat the first result as something to inspect before sharing.
Compare the result with the source at the point where the movement goes wrong. Check whether a limb left the frame, another person crossed in front, or the camera cut to a different shot. Replace a tight photo with a wider crop when needed. Change one input at a time in Motion Transfer AI, then review the same moment again to judge the difference.
Finished results appear in My creations on this page. Play the available clip, inspect the face and moving limbs from beginning to end, then use its download control. Signed-in users can also reopen available results from their history. Download the version you want to keep or share; a preview alone does not tell you whether every movement in the clip is usable.
Choose a motion template or add your own reference, then upload a character photo. Make your first Motion Transfer AI clip and review it in My creations.
