Motion Transfer: Animate a Single Character Photo With a Driving Video
TL;DR
TL;DR: Motion transfer takes one character photo and one driving (performance) video, then maps the driver's body pose, head turns, and facial expressions onto your still image so the character moves, talks, and emotes. On Viralance this runs on ByteDance's DreamActor v2 in video-to-video mode, accepts driving clips up to 30 seconds, and costs 4 credits for a 5-second result, 6 for 10s, 8 for 15s, and 17 for 30s.
What is motion transfer?
Motion transfer is a technique where the motion from a driving video is reenacted by a subject in a source image. You supply two inputs: a photo of one character (a person, an illustrated avatar, a mascot) and a video of someone performing the actions you want — gesturing, speaking, nodding, dancing. The model extracts the driver's skeleton, head orientation, and facial movement, then renders your photo's character doing the same thing while keeping that character's face, clothing, and identity intact.
It is sometimes called pose transfer, character reenactment, or "puppeteering" a photo. The key distinction from plain image-to-video: you are not describing motion in a text prompt and hoping the model invents it. You are giving the model an exact performance to copy.
Related questions: What does motion transfer mean? Is motion transfer the same as deepfake? What is character reenactment?
How does motion transfer differ from image-to-video?
Image-to-video (a model like Luma Ray 2 or Seedance 2.0 animating a still) generates motion from a text prompt. You write "she waves and smiles" and the model improvises the timing and gesture. Results are creative but unpredictable, and you cannot dictate exact choreography or lip timing.
Motion transfer is deterministic about movement. The driving video is the choreography. If the driver tilts their head left on beat three and raises an eyebrow, your character does the same on beat three. That makes it the right tool when you need a specific performance — a precise dance, a synced talking-head delivery, a repeatable brand gesture — rather than a generic "make it move."
Related questions: Should I use image-to-video or motion transfer? Can text-to-video copy a specific dance?
How do you animate a photo on Viralance?
Inside Viralance the feature is called Motion Transfer, powered by DreamActor v2 (fal-ai/bytedance/dreamactor/v2) and listed under video-to-video models. The workflow:
- Upload your character image — one clear subject, framed so the body and face you want animated are visible.
- Upload or pick your driving video — the performance to copy. Input clips up to 30 seconds are supported.
- Generate. DreamActor v2 reads the driver's pose and expression frame by frame and reenacts it on your photo's character.
A practical note from the pipeline: the first second of the driving clip is trimmed by default to stabilize the opening pose, so plan your driver to have a neutral lead-in.
Related questions: How do I make a photo talk on Viralance? What model does Viralance use for motion transfer?
What can the character actually do — talk, emote, dance?
Because the source of motion is a real human performance, the character inherits whatever the driver does within frame:
- Talk — mouth shapes follow the driver's speech movements, so a talking-head driver produces a talking character.
- Emote — eyebrow raises, smiles, blinks, and head tilts carry over, which is what makes the result read as alive rather than a warped photo.
- Gesture and move — upper-body pose, hand raises, leans, and turns transfer when they are visible in the driving clip.
For audio, motion transfer animates the visual performance; pair it with a voice track to land lip-sync. Viralance covers that side separately with Seedance 2.0's lip-sync path and ElevenLabs voices, so you can drive the face here and attach the matching voice in the same project.
Related questions: Can motion transfer make a still image talk? Does it transfer facial expressions? Can I add a voice?
What does motion transfer cost on Viralance?
DreamActor v2 is billed by output length. Credit costs:
- 5 seconds — 4 credits
- 10 seconds — 6 credits
- 15 seconds — 8 credits
- 30 seconds — 17 credits
The underlying rate is roughly $0.05 per second of generated video. Because pricing scales with duration, a common workflow is to test choreography on a 5-second clip (4 credits) before committing to a 30-second render. Viralance uses one-time credit packages with no subscription, so unused credits stay in your balance.
Related questions: How many credits is DreamActor v2? Is motion transfer cheaper than full video generation?
How do you get clean results?
A few input choices do most of the work:
- One clear subject in the photo. A single, well-lit character with an unobstructed face transfers more reliably than a crowded or low-resolution image.
- Match the framing. If the driver is a waist-up talking head, use a waist-up photo. Large mismatches in crop force the model to invent body parts.
- Keep the driver in frame. Movement that leaves the frame or occludes the face has nothing to map onto your character.
- Stay under 30 seconds of driver input and lead with a neutral pose, since the opening second is trimmed.
Related questions: Why does my motion transfer look distorted? What makes a good driving video? What resolution should the source photo be?
Where does motion transfer fit in a content workflow?
It is strongest when you need the same character to deliver many performances: a recurring brand avatar that talks to camera each week, a UGC-style spokesperson built from one product-holding photo, an illustrated mascot that dances to a trend, or a single founder headshot reused across dozens of clips without re-shooting. Film one driving performance, swap the character image, and you have a consistent on-screen presence across a whole content calendar.
On Viralance, Motion Transfer sits alongside text-to-video, image-to-video, lip-sync, and caption tools, so a clip you drive here can move straight into voiceover, subtitles, and publishing to TikTok, Reels, or Shorts.
Related questions: What is motion transfer used for? Can I reuse one avatar across videos? How do I turn a motion-transfer clip into a finished TikTok?
