Martin Scorsese, the Oscar-winning director behind "Taxi Driver" and "Goodfellas," has officially joined Black Forest Labs as an advisor, embracing generative AI as a creative tool for filmmaking. The 83-year-old filmmaker is now using the company's Flux image generation model to create storyboards for upcoming projects, marking a significant moment when one of cinema's most respected auteurs publicly endorses AI as part of the filmmaking process.

Why Is Scorsese Turning to AI for Storyboarding?

For seven decades, Scorsese has faced a persistent challenge in filmmaking: translating the images in his mind into something his cast and crew can understand. Traditional storyboarding is time-consuming, requiring detailed sketches or descriptions that don't always capture the director's vision. Scorsese described this problem directly, explaining that "there's always been this problem of how do you communicate what you see in your head to your cast and crew". With Flux, he found a solution that he called "creatively freeing."

Scorsese

The practical benefits are significant. During pre-production, time directly translates to money. By using Flux to visualize and immediately share storyboards, Scorsese can move faster without sacrificing quality or craft. He noted that "this allowed us to move faster without sacrificing quality or craft" during testing on a recent scene. The tool also reduces wear and tear on production crews who would otherwise spend hours in planning meetings.

How Is Scorsese Using Flux to Reimagine Classic Filmmaking Challenges?

In a video demonstration filmed at his New York office, Scorsese reflected on one of cinema's most technically demanding sequences: the famous Copacabana Steadicam shot in "Goodfellas." That single scene required meticulous planning, with at least 10 to 12 separate segments, each requiring careful consideration of lighting positions and transitions. Scorsese explained that "the sequence of scenes consisted of at least 10 to 12 short segments, and for each of them we had to physically consider the lighting positions and how to transition between scenes".

Scorsese

With a tool like Flux, Scorsese believes such complex sequences could be planned "much much quicker," saving production time and reducing crew burden. The AI model allows him to test multiple visual approaches in minutes rather than hours, then refine them based on feedback. In the demonstration, Scorsese iteratively adjusted generated images, commenting that "it's getting better" as the model refined its output.

What Does This Mean for Cinema's Evolution?

Scorsese has never shied away from embracing new filmmaking technologies. He used 3D cinematography for "Hugo" and pioneered de-aging visual effects in "The Irishman." His perspective on AI reflects this openness. He stated that "film is still a young medium, with a history of only about 125 years. Therefore, we must be open to how film will evolve". For Scorsese, Flux represents the next natural step in cinema's technological progression.

In a statement posted on Black Forest Labs' website, Scorsese emphasized that AI tools like Flux allow him to "clearly and efficiently share what I envision with the creative team, including production designers, art designers, and cinematographers. They can then develop it further and enrich the cinematic intelligence". This framing positions AI not as a replacement for human creativity, but as a bridge between the director's vision and the collaborative work of the entire production team.

Scorsese

Who Is Black Forest Labs and Why Does This Partnership Matter?

Black Forest Labs is a Germany-based company founded in 2024 by individuals who previously worked on Stable Diffusion, one of the most widely used open-source image generation models. The company developed Flux, an image generation model known for superior performance. Flux has already been integrated into Adobe Photoshop, demonstrating its credibility within professional creative workflows.

Robin Rombach, CEO of Black Forest Labs, told The New York Times that Scorsese's involvement represents "a great proof point that this works". Scorsese was introduced to the firm through BroadLight Capital, an investment company that counts his manager Rick Yorn among its co-founders. This partnership signals that generative AI is moving beyond tech enthusiasts and into the hands of established creative professionals.

Robin Rombach, CEO of Black Forest Labs

How Are Other Hollywood Directors Responding to AI?

Scorsese is not alone in embracing AI, though he is among the most prominent. James Cameron, director of "The Terminator" franchise, "Titanic," and the "Avatar" films, sits on the board of Stability AI, the British company behind Stable Diffusion. Peter Jackson, director of "The Lord of the Rings," recently compared AI to "a special effect" at a Cannes masterclass, suggesting openness to the technology.

However, not all filmmakers share this enthusiasm. Steven Spielberg has expressed skepticism, saying he is comfortable with AI as a creative tool but uncomfortable with AI writing scripts. Guillermo del Toro, director of "Pan's Labyrinth," has been more vocal in his criticism, stating he would "rather die" than use generative AI in his films.

Steps to Understanding How AI Storyboarding Works in Professional Filmmaking

• Visual Ideation: Directors input text descriptions or reference images into Flux, which generates multiple visual interpretations of a scene in seconds, allowing rapid exploration of compositional and lighting options.
• Iterative Refinement: Filmmakers adjust prompts and parameters based on generated outputs, gradually refining the AI's output until it matches their creative vision, much like traditional storyboard sketching but at accelerated speed.
• Crew Communication: The generated storyboards serve as reference materials for cinematographers, production designers, and art directors, providing a shared visual language that reduces ambiguity during pre-production planning.
• Time and Budget Savings: By compressing the storyboarding phase, productions reduce pre-production costs and allow crews to move into principal photography with clearer creative direction and fewer planning delays.

What Research Reveals About How Diffusion Models Generate Images

Behind tools like Flux lies sophisticated research into how diffusion models work. Recent advancements reveal that semantic structure in these models doesn't emerge smoothly over time; instead, it transitions sharply from ambiguity to clarity. Researchers have found that by tracking class-conditional entropy, a measure of information uncertainty in the model's latent space, they can identify critical noise regimes where semantic decisions crystallize.

This research has been validated on Stable Diffusion 1.5 and other models, showing that entropy-driven insights can lead to more effective model guidance and control. Understanding these transitions helps researchers refine how AI models generate images, which ultimately improves the quality and reliability of tools like Flux that creative professionals use.

The broader implication is that as researchers better understand the mechanics of diffusion models, tools for filmmakers, designers, and other creative professionals will become more predictable and controllable. This scientific foundation supports the practical benefits Scorsese is already experiencing in his storyboarding workflow.