Face-swap localisation is the process of using AI to replace the visible faces in a finished commercial with new talent that better reflects a particular market, while keeping the rest of the production — performance timing, direction, lighting, edit — exactly as it was shot. It lets a single hero film become several culturally specific variants without reshooting.
Why brands use face-swap localisation
Global advertisers face a recurring trade-off. Run one piece of creative everywhere and lose local relevance, or commission a separate shoot per market and lose budget, time, and consistency. Face-swap localisation sits in between: a brand keeps the creative idea, the script, and the production design across every market, and only changes the people on screen so the campaign reads as locally cast.
The technique is most often used in markets where the lead talent's appearance materially changes how the audience reads the ad — categories like beauty, FMCG, health, family, and food. It is also widely used to extend a campaign into markets that were not in the original shoot plan, where commissioning local production is not commercially viable.
How face-swap localisation works
Modern face-swap pipelines work frame by frame. The original footage is analysed to lock the original talent's head pose, expression, and lighting. A new face — either a real, contracted actor or a fully licensed synthetic likeness — is then rendered into each frame with the same pose, expression, and lighting as the original. The result is a clean integration that preserves the original performance and direction, with the new face inheriting the existing cinematography rather than being composited on top of it.
At Myth Labs the work runs through our proprietary Agent Myth AdLocalise pipeline, which combines face-swap with voice cloning, lip-sync, and visual resynthesis of text and packaging. A typical brief covers 3 to 12 markets and delivers all variants in under five working days.
Face-swap, deepfakes, and consent
Face-swap is the same underlying technology as a deepfake; what differs is the consent and the contract. In a properly run advertising project, every face on screen is either a contracted, paid actor whose agreement explicitly permits AI modification, or a fully licensed synthetic likeness with no real-person basis. The ethical line sits at consent and disclosure, not at the technique.
Brands working in regulated categories should also check that the original talent agreement permits AI modification before committing to face-swap. If the talent contract pre-dates the technology, a written addendum is usually needed. Our team walks clients through the typical contractual checks as part of a face-swap brief — see the ad localisation process for the full sequence.
What changes, and what stays the same
The thing that most surprises clients new to face-swap is how little of the original production actually changes. The lighting setup is the same. The wardrobe is the same. The blocking, the camera moves, the edit, the sound design, the grade — all the same. The performance is the same too: the original actor's eye line, micro-expressions, head tilts, and timing are preserved. What changes is the surface of the face, frame by frame, so that the visible identity reads as the new market's talent. Done well, the only person who can spot a face-swap is the original actor.
That has a knock-on effect on consistency. A traditional multi-market production tends to drift across markets even when every shoot follows the same script: different DPs make different decisions, different locations cast different light, different edits land on different frames. Face-swap localisation eliminates almost all of that drift, because every market shares the same underlying production. The brand voice is identical across every variant.
When face-swap is the right choice
Face-swap localisation is a strong fit when the creative idea travels well, the original master is high quality, and the goal is to deliver consistent global creative that still feels locally cast. It is also useful in research, where brands want to test the same script in multiple markets with a controlled variable — only the talent changes — rather than confounding the test with different productions.
It is a weaker fit when the creative concept itself needs to change per market, or when the original master has unstable lighting and tracking that would make any frame-by-frame replacement visible. In those cases, a partial reshoot or a hybrid approach often produces a better result.
Cost, turnaround, and how to brief it
A typical face-swap localisation brief delivers all market variants in under five working days from the point at which the master, the talent direction, and the regulatory copy are signed off. The cost per market variant is a fraction of a reshoot in the same market — broadly £3,000 to £10,000 per market depending on scope, against £15,000 to £40,000 for a full per-market production. The exact number depends on length, the number of in-vision faces, the amount of on-screen text and packaging that needs reworking, and the complexity of the QC pass.
Briefing a face-swap project well comes down to four things: an approved master at broadcast quality, a clear demographic profile for each market's replacement talent, a confirmed talent agreement that permits AI modification, and the per-market regulatory copy. The cleaner those inputs are, the faster the variants land and the less time is spent in QC reprocessing edge-case shots.
For more on how face-swap fits into a wider AI localisation brief, see our long-form guide on the face-swap localisation process.