The power of AI to significantly speed up film and video production times and reduce costs was a hot topic at this year’s Hong Kong International Film & TV Market (FILMART), organised by the Hong Kong Trade Development Council.
Yet, at the same time, participants recognised the fundamental irreplaceability of human input and need for films to retain a human touch.
The “Gearing Up for AI Opportunities” panel brought together production houses from China and the US, which are already heavily using AI to produce animation blockbusters, like Ne Zha 2, video assets for influencers, such as Mr Beast, advertisements and peer to peer short videos.
Benefits and limitations
For David Carpenter from Malka Media, a US-based digital production agency, AI tools account for about 50% of his production process, including pre-production script writing, editing, storyboarding and pre-visualisation.
He said: “As a director, it really is helpful, allowing us to see what the film is going to look like before we even start shooting. You can easily show it to a client and say 'This is what we want to do' before committing time and money.”
However, he noted AI systems can sometimes be hard to finetune to meet a director’s pre-defined vision.
AI’s power in post-production was emphasised by Liu Baoyu from Heguang Post-Production, whose projects have amassed HK$32 billion (US$4 billion) in box office revenue in only 3 years.
At her studio, AI tools are applied in all aspects of post-production, including editing, creating subtitles, reducing noise, synchronising voice to picture, sampling music and producing visual effects. She related that AI is most useful in generating options for, say, a musical or sound effect, which a director can then build upon or use as a reference.
For Hong Li Animation Studio in Shanghai, AI has mostly been used in the art design phase of projects, such as Ne Zha 2, significantly accelerating the speed of rendering compared to traditional methods and collecting and analysing motion data.
However, to date, AI tools struggle to handle the intricate logic structures of animated special effects, stated the studio’s producer Yu Zhixin.
Massification
The mass use of AI is a focus for the Chinese short-video sharing and production platform Beijing Kuaishou, which began as a company specialising in GIF animations and counts 700 million active monthly users.
In June last year, the platform launched its own AI-driven video generation model ‘Ke Ling’, through which users can easily transform static images into animated videos and generate visuals from simple text instructions.
The success of the model is now prompting the company to pivot into becoming a broader AI-driven creation and sharing platform, said Vice President Liu Zhen.
Jury out on legal issues
Most panellists noted that given the speed at which AI models evolve and become faster and cheaper to use, the technology is destined to move beyond production houses and into the living rooms of non-professionals.
In such a context, questions about the ownership of content produced by AI become urgent, yet most legal frameworks are unable to provide definitive answers, according to legal expert Jihong Chen from Zhong Lun Legal.
He noted the issues extend from the ownership of data used to train AI models through to whether AI-generated outputs are legally protected and who bears the liability for the copyright infringement of AI-produced content. These issues, he noted, are core matters to resolve when AI becomes widely applied beyond the film and video industries.
Will AI take over?
On the ultimate question of whether AI technology could produce a film independently of human input, the panellists were united that such a film would be unlikely to meet humans’ basic expectations.
Said Managing Director at Mei Ah Entertainment Group Jason Li: “The key reason why AI cannot yet replace human creativity is that it lacks humans’ unique experiences and feelings about the society in which they live … Ultimately, AI functions as a supportive tool in the creation and production process, enhancing human creativity rather than replacing it.”
Kuaishou’s Mr Liu chimed in noting that the personal touch provided by human e-commerce influencers is very hard to duplicate: “The emotional value provided by many of our sales influencers is something that machines can’t achieve. Sales staff can deploy things like traditional sales techniques and anecdotes passed down through generations. The AI tool just doesn’t have such individualised ways of saying things.”
While all panellists are keen to deploy AI tools for their efficiency and cost, a final takeaway was provided by producer Ms Liu: “AI can transform labour-intensive tasks into data-driven processes. However, the distinctiveness and creative essence of film remain irreplaceable by AI.”