EN AR

AI performers reach cinemas as Kilmer recreation and Tilly Norwood advance

Tamara khadrah

Key Points

  1. An AI recreation of Val Kilmer stars in western As Deep as the Grave, approved by his estate.
  2. AI performer Tilly Norwood is slated for a feature debut in Misaligned after a backlash.
  3. Visual-effects leaders and AI firms disagree on whether synthetic actors can carry real emotion.

The latest:

An AI-generated version of the late Val Kilmer appears in the western As Deep as the Grave, with his estate signing off on the use, according to the Guardian. Director Coerte Voorhees produced the recreation, which the paper described as rigid and short of Kilmer’s familiar screen presence. A separate AI performer, Tilly Norwood, is set for a feature debut in Misaligned.

Details:

  • The estate’s reasoning: Kilmer’s estate approved the digital recreation by citing the actor’s enthusiasm for new technology that expands storytelling, the Guardian reported. Extended footage released earlier this year showed a thirtysomething version of the actor whose voice and manner did not match his known performances.
  • The Norwood backlash: Tilly Norwood, unveiled last year in an introductory comedy sketch, drew hostile reactions from working actors. Emily Blunt called the synthetic performer really, really scary, the Guardian reported, a response the paper read as evidence of how deeply the industry resists digital performers replacing people.
  • The sceptical view: Tim Webber, chief creative officer at British visual-effects company Framestore, said AI models are trained on the surface of things, the pixels of internet images, and that reaching genuine emotion in a performance remains a long way further down the line.
  • The bullish view: Higgsfield AI claimed recently to have eliminated the uncanny valley, the artificial quality affecting even high-end CGI such as the de-aged Indiana Jones in The Dial of Destiny. The company released two entirely AI-generated features online this year.
  • The production numbers: Hell Grind, a 95-minute action fantasy, and The Cully Hill Boys, a 115-minute caper comedy, completed the equivalent of shooting and post-production in a fortnight and a month respectively, excluding scripting, according to Higgsfield content lead Adil Alimzhanov, who used his own face for Hell Grind’s lead.
  • How prompting works: The Cully Hill Boys was written by Tim Planagan and assembled in 30-second chunks using Seedance 2.5. Character assets are built first as body shots and facial closeups for continuity, then each segment is defined by a roughly 4,000-word prompt covering action, camera movement, lighting and emotion.
  • The technical ceiling: In Hell Grind, released in May, characters could sustain only a single emotion per 15-second shot, the limit of the Seedance 2.0 model used. The later film drew its lead trio from likenesses of streetball player Matt Kiatipis, MMA fighter Israel Adesanya and livestreamer N3on.
  • The director’s workaround: Zack London, who makes AI sci-fi shorts as Gossip Goblin, said models often return over-the-top telenovela-type performances, forcing him to treat the system like a marionette doll and adjust details such as a lip quiver or a shifted gaze. He records dialogue with real actors first.
  • The hybrid camp: Tye Sheridan, co-founder of Wonder Dynamics and star of Ready Player One, said performances should retain a human core. His Flow Studio tool uses AI for cheaper motion capture, while Webber is testing what he calls digital makeup applied to filmed performances.
  • The unanswered question: Sheridan asked whether generative models are designed to deliver legendary performances or to cut to an average. Alimzhanov countered that The Cully Hill Boys prompted the model with contextual information about what characters were experiencing rather than specifying physical mechanics.

Background:

The dispute restages an older argument in acting: the external, formalist reproduction of emotion against the internal method associated with Stanislavsky. The Guardian noted Laurence Olivier’s advice to Dustin Hoffman to try acting rather than stay awake for 72 hours.

Between the lines:

The Guardian argues the formalist approach may now look like the future, since AI reproduces surfaces rather than excavating psyche. London’s account points the same way: output suited to larger-than-life roles rather than subtle ones maps onto blockbuster performances built from simple, repeatable emotional beats. The paper also notes no AI film has yet done for the technology what The Matrix did for CGI, leaving the economic case unproven.

What’s next

Watch the release of Misaligned, Tilly Norwood’s feature debut, and whether As Deep as the Grave draws audiences. Also watch for the first AI-generated film with a theatrical box office showing, testing whether commercial demand exists.

What to read next