AI object swap: replace a person, outfit or product in a video and keep the shot
An object swap changes one thing inside a video you already have and leaves everything else alone. You upload the clip, add a photo of what should be there instead, and write a label saying what it replaces — "the woman's red dress", "the drink can on the table", "the man on the left". The output is the same shot: same camera move, same timing, same background, same everyone else, with the labelled thing replaced by what your photo shows. On VdoBloom this is the Object Swap preset of Genjutsu, at /dashboard/video-creation/genjutsu/.
It is not a face filter or a green-screen trick. The model — ByteDance Seedance 2.5 in its editing mode — re-renders the clip with the change integrated into the motion, so a swapped coat follows the wearer's stride and a swapped product sits in the hand that holds it. Identity is kept: in our own test a footballer's kit became a leather coat and his face, his sprint and his knee-slide were unchanged frame for frame.
You can change several things at once. Each reference photo gets its own label, up to ten per job, so a wardrobe change for two people or a product swap plus a background sign are one generation, not three. The honest limit is crowded scenes: with many similar people the model can occasionally apply a change to someone you did not name, which specific labels and distinct photos reduce.
What this tool does
- An AI object swap re-renders the clip with the labelled thing replaced and everything else preserved — camera, timing, background and every unlabelled person.
- One labelled photo per change, up to ten per job: "@Image1 replaces the man on the left; @Image2 replaces the red car" is how the instruction reaches the model.
- Faces and body motion are preserved by default; a swapped outfit follows the wearer's movement rather than being pasted over it.
- Source clips are 2-30 seconds at 480p-720p; the output matches the clip's length and aspect ratio at 480p, 720p or 1080p.
- Typical jobs: wardrobe changes, product replacement in an existing ad, recasting a character, swapping a prop or a sign.
- Billing is per second of clip on the with-reference-video rate: 15 seconds is 212 credits at 480p, 473 at 720p, 852 at 1080p.
- The change is described by labels and an optional one-line note — no prompt engineering, no masks, no rotoscoping.
How it works
1.Open Genjutsu and choose Object Swap
At /dashboard/video-creation/genjutsu/. The Object Swap card shows a real swap from our tests so you know what to expect.
2.Upload the clip you want to change
2-30 seconds, 480p-720p, MP4 or MOV. The price for this exact clip appears as soon as the file is chosen.
3.Add a photo of each replacement and label it
A product shot, a flat-lay of the outfit, a character portrait. The label says what it replaces: be specific — "the striker in the number 10 shirt" beats "the player".
4.Generate and download
Four to seven minutes. The result lands in My Creations at the clip's original length and ratio, watermark-free on paid credits.
Frequently asked questions
Can AI replace a person in a video?
Yes. Label a photo of the new person with what they replace — "the man on the left" — and the output shows them performing the original movements in the original shot. To change only what someone wears, label the photo as the clothing instead and the face is kept.
Can I change an outfit in a video with AI?
Yes, and it is the most reliable swap in our tests: a flat-lay or product photo of the outfit, labelled "the woman's dress" or "the striker's kit", is applied across every frame with the wearer's face and motion unchanged.
Can I replace a product in an existing ad?
Yes — upload a clean product shot and label it with what it replaces ("the drink can on the table"). The new product inherits the original's position and handling, so a hand that lifted the old can lifts the new one.
Do I need to mask or rotoscope anything?
No. The model locates the target from your label and the clip. Your only inputs are the clip, the photos, the labels and an optional line of guidance.
Will it change things I did not ask for?
It is instructed not to, and in single-target tests nothing else moved. In crowded scenes with several similar people it can occasionally touch an unnamed person; specific labels and distinct reference photos are the fix, and a second run is cheap at 480p.
What does an object swap cost?
Per second of clip, at the output resolution: a 15-second clip is 212 credits at 480p, 473 at 720p or 852 at 1080p. The figure for your clip is shown before you confirm and refunded automatically if the job fails.
Related tools
Genjutsu AI video: change what is in a clip without reshooting it
Upload a 2-30s clip, label a photo with what it replaces, get the same shot back changed. Faces, motion and camera stay. Seedance 2.5, up to 1080p.
AI motion transfer: keep the people and the moves, change the world around them
Keep every person and movement from your clip and rebuild the location from photos. Faces, poses and camera stay. 2-30s clips, up to 1080p.
Higgsfield Genjutsu alternative: the same engine with labelled targets and credits that don't expire
VdoBloom vs Higgsfield Genjutsu: same Seedance 2.5 engine. Differences: per-photo labels, price before upload, no plan gate, credits that never expire.
Seedance 2.5: the 30-second model, and how to run it
ByteDance Seedance 2.5 renders 4-30 seconds at 480p, 720p or 1080p (10-bit). Pick it in Text to Video, Image to Video or Effects. 30s at 480p is 276 credits.
Ready to try it?
Swap something in a clip