Video Character Replace
Generated on PixelDojo with WAN 2.2 Replace. Produced by PixelDojo's generation pipeline.
You give the model two things: the video you already have and one still image of the character you want in it. WAN 2.2 Replace then rebuilds the clip with the new character in the old one's place, keeping the background, the framing and the motion. We ran it on August 26, 2026. It charged 10 credits for a 5 second clip and finished in 168 seconds. This page shows what went in, what came out, and where the trade offs sit.
The finished swap, one run, nothing retouched
Every example below was produced on PixelDojo. Hover to see the prompt.
Source clip of a man mending a net at a harbour, character still of a woman with curly red hair, round glasses, denim jacket and yellow shirt, swapped at the 480 tier for 10 credits
WAN 2.2 Replace
What the swap kept and what it changed
The two inputs
Source video: a 5.04 second vertical clip of a man sitting at a harbour at sunset, working a net with both hands, a moored wooden boat behind him. Character still: a woman with short curly red hair, round glasses, a denim jacket over a yellow shirt. Nothing else, no prompt required.
What came out
A 4.97 second clip at 512 by 768, 149 frames, about 2.9 MB. We pulled a frame from the middle of it. She is in his seat, hands on the net in the same pose, and her red curls, round glasses, denim jacket and yellow shirt all carried across from the still.
The scene did not move
Same harbour, same moored boat in the same spot on the right, same low sun on the water, same warm backlight through the hair. This is a replace, not a regenerate, so you keep the shot you already liked.
It is a whole character, not just a face
People search for this as a face swap, but the model brings the hair, the glasses and the clothing from the still as well. If you only want the face changed, feed it a still that matches the outfit already in the shot.
Two resolution tiers, same price
There is a 480 tier and a 720 tier and both charge the same 2 credits per second. We used 480 and a 784 by 1176 source came back at 512 by 768, so the cheaper tier costs you resolution rather than credits. Use 720 unless you have a reason not to.
The source audio can ride along
Keeping the original audio track is a setting on the job and it is on by default, so dialogue and room tone from the clip you started with do not have to be rebuilt afterwards.
One 5 second clip plus one character still. 10 credits, 168 seconds, and the harbour, the boat and the sunset all survived the swap.
Why Choose Pixel Dojo for Video Character Replace
Professional-quality results with cutting-edge AI technology
10 credits for 5 seconds
2 credits per second of output, rounded up to the next second. Clips run from 1 to 15 seconds, so the ceiling on a single job is 30 credits.
168 seconds, timed
Measured on August 26, 2026 from submitting the job to the finished file. Queue load moves it, so read that as typical rather than guaranteed.
No prompt writing
The job takes a video and an image. The performance comes from the source clip, so you are casting a shot rather than describing one.
How It Works
Three steps, which is the whole job:
Pick the shot, not the description
The output copies the source clip's framing, camera move and body motion. Choose footage where the action is already what you want, because the swap will not change it.
Add one character still
Face visible, outfit readable. Ours was a single portrait of a woman with curly red hair, round glasses and a denim jacket, and all three showed up in the result.
Set the length and run
The duration you set is what you pay for at 2 credits per second. Our 5 second job charged 10 credits and returned in 168 seconds.
Loved by creators on PixelDojo
Real feedback from people using PixelDojo, pulled from our in-product surveys.
Prompt updates, strong features, and a community that is always willing to help and hear/act on professional feedback. I've never had a better experience.
user friendly all in one thing
I really like the schedule of AI work performed!
The best tools for IA on the web !
The guy that operates the website is constantly updating it
I’ve been using PixelDojo for around 8 months now and it’s always worked wonderfully while constantly adding features.
Explore more AI tools on PixelDojo
AI Tools
Compare & Switch
- Best AI Image Generators
- Best AI Video Generators
- Midjourney Alternatives
- Civitai Alternatives
- Runway Alternatives
- Leonardo Alternatives
- Pika Alternatives
- Luma Alternatives
- Magnific Alternatives
- Veo Alternatives
- Flux Alternatives
- Freepik Alternatives
- Seedance Alternatives
- Seedream Alternatives
- Pixverse Alternatives
- GPT Image Alternatives
- Synthesia Alternatives
- Playground Alternatives
- NightCafe Alternatives
- Canva AI Alternatives
- ElevenLabs Alternatives
- ComfyUI Alternatives
- Fal Alternatives
- Replicate Alternatives
Common Questions
Everything you need to know about Video Character Replace
How do I swap a character in a video with AI?
Upload the source clip, upload one still of the character you want, set the clip length and run it. The model replaces the person in the footage while holding the background and the motion. Ours took 168 seconds for a 5 second clip on August 26, 2026.
What does a video character swap cost?
2 credits per second of output, rounded up. Our 5 second run charged 10 credits. Clips can be 1 to 15 seconds, so one job tops out at 30 credits.
Is this a face swap or a full character swap?
A full character swap. In our run the new character brought her hair, glasses and clothing into the shot, not just her face. To change the face alone, use a still that already matches the wardrobe in the footage.
What resolution do I get back?
There are 480 and 720 tiers and they cost the same per second. We ran the 480 tier and a 784 by 1176 source returned at 512 by 768, so the lower tier trades pixels rather than credits.
What makes a good character image?
One still with the face clearly visible and the outfit readable. The model carries what it can see, so a small, dark or heavily cropped still gives it less to work with.
Can I call it from the API?
Yes. POST to /api/v1/models/wan-2.2-replace/run with your API key, the source video URL, the character image URL and the clip duration. Same key and same request shape as every other model in the catalog.