Behind the scenes of TV's first deepfake sketch comedy
ITV is rolling out Deep Fake Neighbour Wars, a new comedy using AI face-swapping from StudioNeural to turn everyday neighbor bickering into celebrity drama. Actors had to keep strictly facing the camera to avoid breaking the synthetic effect. The show hits ITVX on January 26.
Imagine turning on your TV and seeing multimillionaire superhero Tom Holland living in a cramped flat, whining endlessly about his petty living situation. It looks completely real, sounds identical, and feels deeply confusing.
That is the premise behind Deep Fake Neighbour Wars, a new sketch comedy series premiering January 26 on the UK’s ITVX, pitched as television’s first full-length deepfake comedy.
Everyday drama for A-list faces
The setup is simple: massive celebrities live as ordinary working-class neighbors, constantly bickering over mundane domestic disputes. Think England striker Harry Kane complaining that grime artist Stormzy cracked his patio tile, or Mark Wahlberg stripped of his fitness obsession and dealing with asthma.
Behind the tech is Barney Francis, head of StudioNeural, who prefers the term synthetic media over deepfake. His team trained models over millions of iterations to generate hyper-realistic celebrity faces, mapping them onto up-and-coming impressionists.
Lead writer Spencer Jones loved leaning into the absurdity. Rather than shooting for polished perfection, the production deliberately paired recognizable Hollywood faces with mundane, slightly beaten-down body types.
Making TV without turning your head
Pulling this off inside an actual broadcast workflow revealed how fragile deepfake technology still is.
Because the tracking algorithms require consistent facial landmarks, actors were forced to face the camera constantly. If a performer naturally turned sideways to make eye contact with their scene partner, the synthetic mask would simply collapse.
Actor Katia Kvinge noted that filming felt like being physically restricted, constantly fighting the urge to move naturally. Physical occlusions were strictly off-limits: no looking out through rainy windows, no touching the face, and no wildly exaggerated expressions.
Casting also relied heavily on underlying facial architecture. Kvinge only landed the role of Olivia Colman because her natural facial geometry matched the Oscar-winner by 95 percent.
Why it matters
For years, deepfakes have lived in the public imagination primarily as tools for political disinformation, fraud, and non-consensual exploitation. This show marks the first time a mainstream broadcaster is pushing synthetic media straight into prime-time living room entertainment.
For production companies, the upside is clear. High-profile celebrity cameos that would normally cost millions in talent fees can now be simulated using affordable impressionists and GPU hours, opening up a wild new creative sandbox for sketch comedy and satire.
That said, we think the legal and ethical ice here is remarkably thin. ITV is attempting to play it safe by slapping disclaimers at the start of every episode and letting the real actors unmask at the end. The creators argue it is harmless parody with zero malice, but widespread broadcast use of unauthorized digital likenesses is guaranteed to trigger pushback from talent unions and estate lawyers.
Technically, the show also proves that fully automated AI filmmaking is nowhere near ready. Squeezing high-definition facial textures onto moving bodies still demands enormous human supervision, highly staged shooting conditions, and endless manual tweaking. It works for a goofy, locked-off sitcom sketch, but real cinematic flexibility remains years away.
Sources (1)
- theguardian.com(opens in a new tab)Originaltheguardian.com