AI is changing how directors and cinematographers work—but not the way you might think

AI is changing how directors and cinematographers work—but not the way you might think

When people think of artificial intelligence in Hollywood, they might picture deepfakes, synthetic actors, or AI-generated scripts and video. Google’s Veo3, along with other tools like Pika Labs and Kling AI, made headlines for their photorealistic AI generated video clips (as did OpenAI’s Sora 2 before the company in March announced plans to shutter it). 

But for freelance filmmakers, the real shift is happening behind the scenes. For years, cinematographers and directors have had to wear many hats: artist, technician, project manager, negotiator. Now, AI is quietly taking over some of the more tedious jobs. 

Short-Form Frontier

Michael Goi, former president of the American Society of Cinematographers (ASC) and current co-chair of its AI committee, remembers widespread panic in the industry a few years ago. “There was this blanket fear that AI would completely replace jobs,” he says. 

That fear has been overblown, Goi says. He presented an ASC seminar last year outlining one of the largest hurdles to widespread adoption of AI video—consistency. In a live demonstration with six-time Oscar-nominated cinematographer Caleb Deschanel and AI creator Ellenor Argyropoulos, the filmmakers attempted to use AI tools to generate a specific shot. “Caleb had a very clear vision,” says Goi, “and it was a struggle to even get close.”

Though video AI tools have made significant strides since then, they are still very much geared toward short-form content, with most tools only capable of generating clips of up to two minutes in 4K quality. That’s good news for the growing number of people working on vertical series—Goi among them—who get to test new video-generation models, sometimes before their public launch.

A striking example of what’s now possible is Fruit Love Island, an AI-generated “fruit slop” microdrama from TikTok account @ai.cinema021 that became the platform’s fastest-growing account ever, amassing over 3 million followers in nine days and 300 million total views before coming to an abrupt halt in late March after being flagged for low quality. Each two minute episode allegedly took around 3 hours to make, and are thought to have used text-to-script tools like Object Talk that are then plugged into an AI video generator.    

For most freelance cinematographers, though, the gains of AI aren’t on-screen, but behind the scenes, making it easier to plan how they will capture the shots they need.    

Streamlining storyboards

While fully AI-generated feature films may not be around the corner, filmmakers are regularly using tools like Midjourney and Runway to create storyboards and visual references. 

Rob Berry, a freelance cinematographer whose clients include Bergdorf Goodman and Nordstrom, Berry remembers his first encounter with AI-generated storyboards on a commercial project. “[The clients] were able to make them very quickly, change them the day before the shoot and hand them to me. I was like wow, the future’s here,” he says. 

Director Sage Bennett, who’s shot campaigns for Dior and Jim Beam, sees a similar trend. “Budgets are getting smaller, and expectations are getting bigger,” she says. In her experience, AI is often being used to bridge that gap, though it still needs a human touch. While last year she still thought AI visuals looked a bit “uncanny valley,” she thinks the technology has gotten much better, and she now sees it as almost standard practice for storyboarding and generating visual references. 

Both Berry and Bennett use AI as a kind of creative sounding board: one that never needs to sleep. “Sometimes you just need to talk through a tiny idea,” Sage says. “I’ll ask, ‘Should I push in or pull out for this shot, and why?’ It helps me sharpen my instincts.” Goi also has used AI to suggest focal length or composition for a shot after plugging in a storyboard. 

Berry says AI doesn’t come up with the ideas, but it’s a great tool for organizing his thoughts in pre-production. Both say that in projects that they’ve been on, AI has mostly been used for voiceover or VFX work rather than production itself. On one commercial, Bennett’s team used an AI-generated voiceover as a placeholder while they sourced a real actor — and ended up preferring the AI for the final product.

Even Steven Soderbergh has leaned in: in a recent interview with Variety, the director revealed he used AI-generated imagery in his John Lennon documentary to visualize surrealist sequences that would have otherwise been out of budget with a VFX house. “My job is to deliver a good movie, period,” he told Variety. “And this tool showed up at a moment when I needed it.”

An Invisible Assistant

Where AI shines most for freelancers like Berry is in logistics. “As a creative freelancer, I’m first and foremost running my own business,” he says. He uses tools like ChatGPT to manage his workload: drafting emails, balancing budgets, and organizing project notes. “I told ChatGPT to act like it was my agent at CAA and walk me through a negotiation,” he says. With seven projects on his plate, he says, “If something could scan my inbox and tell me where I’m at with each one, that’s the dream.” 

Bennett also uses AI to streamline pre-production tasks. “I’ll plug in script notes with descriptions, shot sizes, and ask ChatGPT to generate a clean shot list that I can then go in and adjust. I’ll still tweak it, but it saves so much time.” When writing treatments to pitch commercial work, she sometimes uses AI to help with structure and polish. “I still revise everything in my voice, but it speeds up the process.” 

Though companies have begun testing AI generated commercials, Rob hasn’t seen work slow down for him. But he sees staying on top of AI as part of the job now. He’s been teaching himself prompt engineering through hours of trial and error. “Most people ask a question, get an answer, and leave. But if you keep probing and try different characters and approaches, you get way more out of it.” He particularly likes the “deep research” feature of ChatGPT for in-depth reports on, for example, deciding between two cameras, or developing a pre-production checklist for an ASC-level Director of Photography. “It takes a few minutes and comes back with a ten page report, 16 sources.”  He believes that being adept at using the latest technology is key to staying at the forefront of his craft. 

Goi agrees. “There are conversations I’ve had with Jim Cameron and Rob Legato that AI won’t make a mediocre filmmaker great,” he says. “But it can help a great one refine their vision. That’s why we need top of the line filmmakers involved in where this tech is going. The more professionals engage in what should be best practices for [AI’s] use in the industry, the better positioned the technology and creative artists will be as we progress.”

View the full article