The New Black AI now swaps the model in a fashion video
Published 18 September 2026
A campaign shot with one model can now become a campaign with several. The New Black AI swaps the person in a video: give the clip and a photo of the new model, and the video comes back with that person in it: same outfit, same walk, same setting, same camera.
What Swap Model in Video does
Swap Models does it for a photo; this does it for a clip. The outfit, the movement, the background and the camera stay exactly as filmed. Only the person changes, and the new face and body move with the footage frame after frame.
How it works
1. Upload the video: a Model 360, a runway walk, a UGC clip, your own footage. 2. Add a photo of the new model. A clear photo, face visible, front on. A Product to Model creation or a portrait from your files both work. 3. Generate. The clip keeps its length, its shape and its sound.
What it costs
By the second of the video you hand in: 2.8 credits per second in Standard, 3.7 in Pro. A 5-second clip is 14 credits in Standard.
Why it matters for a brand
Diversity in a campaign usually means a second day of shooting. Here it means a second photo. The same walk can be shown on several models for several markets, and a store can show a garment on the body its customers will recognise.
The AI models generated in the studio are a natural source for the new face. Details on the Swap Model in Video feature page.
Frequently asked questions
What photo of the new model do I need?+
A clear photo, face visible, front on.
Does the outfit stay the same?+
Yes. The garment, the movement, the setting and the camera are kept; only the person changes.
How long can the clip be?+
The clip is priced by the second, and its length is read from the file before the job starts.
Related articles
- The New Black AI now removes the background of a fashion videoA clip comes back as the subject alone: transparent for your editor, white for a product page, black or green for a composite: with hair, fabric and shadows handled by the model, no green screen.
- The New Black AI now edits a fashion video from a single sentenceUpload a clip, write what should change: make it night, add snow, put the model on a beach: and get the same video with that one change. Priced by the second.
- The New Black AI Now Creates Virtual Try-On Videos for Fashion BrandsYour garment on the model you supply, in motion: the clip that answers the question a still cannot, whether the piece holds when the person moves.
- The New Black AI now changes the outfit in a fashion videoMake the jacket red, change the print, replace the top: the clothing in your video changes with a sentence and the person, movement and background stay as filmed.
- The New Black AI Now Turns a Model Photo into a Runway Walk VideoOne photo of your model, one sentence about the movement, and the look walks. From 5 to 30 seconds, so a whole line can move instead of three pieces.
- The New Black AI Now Creates Model 360 Videos from a Front and a Back PhotoThe model turns and the outfit shows all around: length, volume and how the piece sits, which is what a size chart cannot carry.
- AI Fashion Video From Your Photos: UGC Clips, Model Walks and Product 360 TurntablesThree guided video workflows from your photos: a UGC clip with a presenter and voice, a runway walk from one model image, a turntable from the front and the back. What each needs, costs and is for.
- The New Black AI Now Renders Fashion Videos in 720p or 1080p with Generated AudioResolution and Audio sit in the panel of every video workflow: 720p or 1080p, and a voice, a music or an ambience described in words and generated with the clip.
- The New Black AI now changes the background of a fashion videoKeep the model walking, turning or speaking exactly as filmed, and change the world around them to a Paris street, a white studio or a rooftop at dusk.
- The New Black AI now reframes a fashion video to vertical, square or 4:5 without croppingOne clip, five channels: 9:16 for Reels and TikTok, 1:1 for the feed, 4:5 for the grid, 16:9 for the site. The subject is recentred and the missing edges are generated.