AI ‘upgrades’ to kung fu classics deserve zero stars
By Jason Bailey
WHEN it comes to artificial intelligence (AI), most folks seem to fall into two diametrically opposed camps, to . There are those who believe it to be nothing less than a technological revolution that will significantly change every aspect of how we live, and others who dismiss it as the emperor’s new clothes, an ecologically irresponsible, not-ready-for-prime-time pipe dream that produces error-ridden prose and eye-sore attempts at illustration.
China Film Foundation and its partners’ of the Kung Fu Film Heritage Project reminded me why I’m in the latter group.
There are certain, specific, mostly academic and scientific instances in which AI can reduce mindless busywork and produce quicker results. But this technology, particularly its “generative” subset, should be kept far, far away from anything resembling art.
China Film Foundation’s plans, unveiled at the Shanghai International Film Festival, feature two major AI-driven initiatives. First, the organization premiered what was promoted as the first animated feature film created entirely by AI called A Better Tomorrow: Cyber Border. It’s a reimagining of John Woo’s groundbreaking 1986 feature, A Better Tomorrow. Why anyone would want to see an AI sequel/remake to a nearly 40-year-old action classic is beyond me, that’s their choice though. But Cyber Border remains worrisome in the same way as all the gen-AI moviemaking we keep hearing about: because it will replace human writers, actors, artists, and other craftspeople with automated, soulless junk machines.
These are questions that are being actively argued, with real implications for the future of motion pictures, and we’re nowhere close to answers. The more troubling piece of the foundation’s Kung Fu Film Heritage Project is its intention to use AI to upgrade the audio, visuals and overall production of 100 classic kung fu movies (including titles by such legendary performers as Bruce Lee, Jackie Chan, and Jet Li). Zhang Pimin, chairman of the China Film Foundation, that its AI tweaks to these “aesthetic historical treasures” would enhance them with a new look that “conforms to contemporary film viewing.” And that’s where the alarm bells should go off, if you care at all about the preservation and longevity of movies.
This type of automated restoration is not altogether unheard of. Last spring, film purists were over the long-awaited HD restorations and releases of James Cameron’s films Aliens, True Lies, and The Abyss, arguing that the extensive clean-up work on these decades-old movies had rendered the images sterile and often uncanny. Much of that work was done via AI by filmmaker Peter Jackson’s company, Park Road Post Production. It also did similarly controversial clean-up work on Jackson’s archival documentaries The Beatles: Get Back and They Shall Not Grow Old using the same proprietary machine-learning software.
Keep in mind, Cameron’s “upgrades” were performed on films that were already slick, big-budget, Hollywood productions — and the results still looked off. Applying similarly transformative technology to the Eastern action cinema of an earlier era, where budgets were significantly lower and the look and feel of the films was consequently grubbier, would be artistically catastrophic. The great kung fu filmmakers, such as Chan, Lee, and their early director Lo Wei, had distinctive styles. The handmade quality of their pictures would, it seems safe to assume, be flattened and homogenized by these AI overhauls.
But what’s most upsetting is that it’s all so unnecessary. I’ve seen vintage kung fu movies in theatrical settings in recent years, including the national re-release of Chan’s Police Story series and the monthly kung fu double features at Los Angeles’ New Beverly Cinema, and these are not films that struggle to connect with contemporary audiences. Decades after their production, they play like gangbusters, mostly because mainstream action movies are still echoing their ferocious energy and rhythms. The idea that these fast-paced, action-packed efforts need new, fancy, digital bells and whistles to connect with 21st century viewers is patently absurd.
I’m reminded of another computer-powered technological advance, all the way back in the 1980s, that we were assured would make “old movies” relevant again: colorization. To achieve it, computer software was used to slap a layer of color over black and white movies, including Topper, It’s a Wonderful Life, and . These new versions were blasted by critics (most notably Gene Siskel and Roger Ebert) and filmmakers, including Orson Welles, who was able to use the creative control clause of his original RKO contract to halt the colorization of Citizen Kane. But media figures like Ted Turner saw colorization as a way to squeeze more money out of their existing libraries — or, to put it in contemporary terms, to further monetize on IPs.
Alas, colorization flopped with consumers. Ebert gleefully reported in his book Questions for the Movie Answer Man that “the national VHS tape sales of Turner’s 1989 colorized version of Casablanca totaled — get this — less than 600 copies.” It failed for the same reason I suspect these AI-powered restorations will: because their most likely audience will only be annoyed by the so-called improvements, and potential new, young viewers aren’t going to be fooled by this shoddy technology.
Respectful restorations and theatrical re-releases , because great films are timeless, and cannot be improved upon by reckless applications of the flavor of the month. It’s ironic, really, that the Kung Fu Film Heritage Project sees itself as honoring the classics when all it is promising is filmmaking vandalism. — Bloomberg Opinion
