Digital Domain's Charlatan is proprietary, not licensed out

The shot

Digital Domain describes Charlatan, on its own technology page, as growing out of its Digital Human Group work on 'The Curious Case of Benjamin Button' and culminating in 'the studio's proprietary new face-swapping technology called Charlatan.' The page states 'Charlatan leverages machine learning's ability to convert one image to another,' and names two productions: for Malaria No More's 'Malaria Must Die' campaign, the studio 'added nearly 30 years to David Beckham to deliver a speech in the future,' merging current footage with an older stand-in, and for a posthumous performance by singer Teresa Teng, Charlatan handled 'facial transfer seamlessly' in real time. A separate Digital Domain news post, dated 8 February 2021, describes a third use: a 'digital human double' of coach Vince Lombardi built for the NFL's Super Bowl LV broadcast with agency 72andSunny, tied by the post's own headline to Charlatan.

What the documents show

Both documents are Digital Domain's own published materials, describing its own tool and client work; they are not independently verified benchmarks, and the claim that Charlatan 'leverages machine learning's ability to convert one image to another' is the studio's own characterization, not a published specification. The Lombardi post is thinner on method: its body text describes the result, 'the digital return of NFL legend Vince Lombardi' reaching 'over 100 million fans,' without repeating the word Charlatan, so the link rests on the headline and URL rather than a technical description. Neither document states Charlatan is licensed or sold to other studios; it is described throughout as an in-house capability.

The workflow

The technology page describes two applications built on the same image-conversion approach: digital aging, projecting a performer forward in age using a stand-in as a base, and real-time facial transfer, changing a performer's on-screen appearance live via facial tracking. Both, on the studio's account, required merging or tracking footage of a real performance rather than generating a face from nothing. Which approach fits a given production is a decision the studio and client make per project; the page does not claim one replaced the other.

What the tool does not change

Digital Domain's own materials describe Charlatan as built on an existing performance, not a system that invents a likeness or dialogue independently; a real performer's blocking and delivery still drive what the tool merges or transfers. This is editorial: a proprietary in-house tool shown on a handful of campaigns is a narrower claim than a commercial product.

• Does a studio's technology page name specific productions, or describe a capability only in general terms?
• Is a named tool described as licensed to other studios, or used exclusively in-house on the studio's own client work?
• Does the underlying performance in a de-aging or face-transfer shot come from the credited actor, a stand-in, or archival footage?

Charlatan is a documented example of a facility keeping a machine-learning tool in-house and pointing to named campaigns as evidence rather than publishing it as a product; reading the studio's own page for what it claims, proprietary, in-house, performance-based, is more accurate than assuming a demonstrated capability is a marketed one.