WAN 2.2 Animate
Transfer body movement and facial expression from a driving video onto a character image. 480p or 720p, 5–120s, from 10 credits per second.
WAN 2.2 Animate copies motion from a driving video onto a character image. Upload a clear picture of the character and a clip whose movement you want to reuse; the model transfers body pose, gestures and facial expression while keeping the character's identity. Two modes cover different jobs: animate brings the image to life, and replace swaps the character into the original footage. Billing is 10 credits per second at 480p and 30 at 720p.
- Input: Image + video
- Modes: Animate / replace
- Price: 10–30 credits/s
- Resolution: 480p / 720p
How to use WAN 2.2 Animate
Match the character image to the driving footage before anything else. The transfer only works when the source and the motion agree.
- Choose the mode Use Animate when the character image should perform the motion in a new frame. Use Replace when the character should appear inside the original footage.
- Upload a character image that matches the framing A full-body driving shot needs a full-body character image. Mixing a headshot with a walking clip forces the model to invent the body.
- Upload the driving video Keep it to one person and clear motion. The model copies pose, gestures and expression, so busy group footage produces confused results.
- Add optional guidance A short prompt can protect details: preserve the outfit, keep the face neutral, maintain the hairstyle. Long prompts are not needed.
- Pick resolution and watch the billable clamp 480p is the cheap default; 720p costs three times as much. Billing runs from 5 to 120 seconds even when the source is longer.
Settings reference
WAN 2.2 Animate transfers motion from a driving video onto a character image. Two modes cover animation and replacement.
| Setting | Value | Notes |
|---|---|---|
| Character image | 1 image, required | A clear, full or upper-body shot of the character you want to animate. |
| Driving video | Required, up to 100 MB | Source footage up to 300 seconds; the model copies body movement and facial expression. |
| Mode | Animate / Replace | Animate brings the image to life; Replace swaps the character into the original footage. |
| Resolution | 480p / 720p | 10 credits per second at 480p, 30 at 720p. 480p is the default. |
| Prompt | Optional | Short guidance such as preserving the outfit or keeping a natural expression. |
| Seed | Optional | Reuse a seed to reproduce a take while adjusting guidance. |
| Billing | 5s – 120s clamp | A 300-second driving video still bills at most 120 seconds. |
Animate versus replace
Animate mode ignores the driving video's appearance and uses only its motion. The output is your character, in the image's setting, performing the movement from the clip. That is the right mode for bringing a portrait or illustration to life, or for making a mascot dance without rebuilding its world.
Replace mode keeps the driving video's scene and swaps the person in it for your character. It is closer to a visual effect: the background, camera and timing stay, while the performer changes. Replace is more sensitive to the source footage, because the character must fit the original body shape, lighting and occlusion.
- Animate: character and image background, copied motion
- Replace: original scene, substituted character
- Both keep the driving clip's timing
- Replace needs a closer match in pose and light
- Optional prompt for guidance
Pricing by resolution and clip length
Billing follows the driving clip's length. 480p costs 10 credits per second; 720p costs 30. A 5-second clip is 50 or 150 credits; a 30-second clip is 300 or 900. The create page is designed around clips from 5 to 120 seconds, which covers most social and presentation use cases.
Because 720p is three times the 480p rate, prototype at 480p until the motion and identity look right. A 120-second 720p run costs 3,600 credits, so it is worth confirming the take before committing.
- 10 credits per second at 480p
- 30 credits per second at 720p
- Designed for 5–120 second clips
- Uploads up to five minutes are accepted
- No prompt required
| Clip length | 480p | 720p |
|---|---|---|
| 5s | 50 | 150 |
| 15s | 150 | 450 |
| 30s | 300 | 900 |
| 60s | 600 | 1,800 |
| 120s | 1,200 | 3,600 |
Preparing the character image
Use a full-body or at least knee-up image with a clear silhouette, even lighting and no heavy occlusion. The model needs to see the limbs it is going to move. A face-only headshot cannot support a walking or dancing performance, because there is no body to animate.
Match the image's framing to the driving video's framing. If the video is a wide shot of a dancer, a tight portrait will force an awkward composite. A neutral pose with arms slightly away from the body gives the model more room to reproduce the target motion than a folded-arms stance.
- Full-body or knee-up framing
- Clear silhouette and visible limbs
- Even lighting, minimal occlusion
- Neutral starting pose
- Match framing to the driving clip
Choosing and trimming the driving video
The driving video supplies timing, so pick one with a single continuous performance: a dance phrase, a walk cycle, a gesture or a head turn. Cuts and camera changes inside the clip confuse the transfer. Trim to the section you actually need before uploading; there is no reason to pay for seconds you will discard.
For replace mode, choose footage with one clearly visible performer and a relatively stable camera. Fast occlusion, crowds and heavy motion blur make substitution harder. For animate mode, the background of the driving clip is irrelevant, so a plain studio reference works just as well as a location shot.
- One continuous performance, no cuts
- Trim to the needed section
- One visible performer for replace
- Stable camera and light
- Plain backgrounds are fine for animate
Prompting and limits
The prompt is optional and should be used for constraints, not description: preserve outfit, natural expression, no background change. It is most useful in replace mode, where you may need to keep a costume or avoid altering the scene. In animate mode, the image already defines the look.
The endpoint stops at 720p, produces no audio and has no explicit last-frame control. For a still-led cinematic shot without a motion source, HappyHorse Image-to-Video gives you 1080p and a 3–15 second range. For a performance copied from video, Animate is the specialised tool.
- Optional guidance prompt
- 720p maximum
- No audio
- No last-frame control
- Use HappyHorse I2V for cinematic still animation
Prompt examples for WAN 2.2 Animate
- Dance performance: Preserve the character's outfit and face, natural expression, no background change. Use a full-body character image.
- Walk cycle: Keep the outfit consistent, natural arm swing, stable framing, no scene change. Trim the driving clip to one stride cycle.
- Character replacement: Replace the performer with the reference character, preserve the original lighting and camera, keep body proportions natural. Match the source performer's build.
- Gesture transfer: Preserve identity, natural hand movement, no extra objects, keep the original framing. Avoid clips with heavy hand occlusion.
- Facial performance: Keep the face consistent, subtle expression changes, no head deformation, natural blink timing. A clear, front-facing image works best.
- Costume consistency: Preserve the red coat exactly, no colour shift, keep the original scene and camera motion in replace mode. State the garment explicitly.
Pro tips for WAN 2.2 Animate
Motion transfer lives or dies on the source pair. Choose footage and images that already belong together.
- Animate vs Replace Animate generates a new scene around the character. Replace keeps the original footage and swaps the performer. Pick by whether the background should survive.
- Front-facing, unobstructed faces The model needs to see the face to transfer expression. Hair, hands and props across the face break the illusion.
- One performer in the driving video Multiple people in frame make the model average their motion. Trim the clip to the person whose movement you want.
- The 120-second billing cap You can upload up to 300 seconds, but billing stops at 120. Long source clips are fine; just expect a capped charge.
- Deliver at 720p 480p is useful for checking whether the transfer works. Final character work usually needs 720p to hold up on a phone screen.
Frequently asked questions
How much does WAN 2.2 Animate cost?
10 credits per second at 480p and 30 credits per second at 720p, billed on the driving clip's length.
What is the difference between animate and replace?
Animate puts your character into the image's own setting and copies the motion. Replace keeps the driving video's scene and substitutes your character into it.
How long can the driving video be?
Uploads up to five minutes are accepted, but the workflow is designed around 5–120 second clips. Longer inputs cost proportionally more.
Is a prompt required?
No. A short guidance prompt is optional and is most useful for preserving outfit or background in replace mode.
What resolutions are supported?
480p and 720p.
Does it generate audio?
No. The model transfers motion only; there is no audio control. Add sound in post.
What kind of image should I upload?
A clear full-body or knee-up image with visible limbs, even lighting and a neutral pose.
How does it differ from HappyHorse Image-to-Video?
HappyHorse animates a still with prompt-described motion at up to 1080p. WAN 2.2 Animate copies motion from an actual video and supports character replacement.
Compare with other animation models
- WAN 2.2 Speech-to-Video
- WAN 2.2 Image-to-Video
- Kling Image-to-Video
- HappyHorse 1.0 Image-to-Video
- WAN 2.2 Video Edit
- Browse all AI video models
Specifications
- Provider: Wavespeed
- Model type: Video to video
- Credit pricing: 480p: 10 credits · 720p: 30 credits
- Capabilities: popular, cinematic, realistic
About Conut.ai
Conut.ai is an AI image, video and audio generation platform with 50+ creative tools. It supports text-to-image, image-to-video, face swap, motion control, relighting, inpainting and video effects through models including Nano Banana Pro, Seedream, GPT Image, Sora 2, Kling, WAN and Seedance. Start free — no credit card required.