How LUMA’s sweet dreams were made from this

Can you tell us how the concept for the Dream Brief came into being?

Caroline Ingeborn: We saw the tech industry’s initial message to the creative community was off. Telling creatives they needed to become "prompt engineers" was the wrong chord. We didn’t want to make our platform about the AI; we wanted to make it about the idea and how to bring it to life.

The challenge was that most generative AI video is garbage and everybody hates it. 

During a conversation with Jason Kreher at DE-YAN, we hit on a universal truth: every great creative has a "bottom drawer" filled with incredible, ambitious ideas that were pitched but ultimately killed—usually because of budget, time, or resource constraints. Dream Brief encapsulates how we view Luma’s role in the industry: we are here to ensure that complex, ambitious ideas can finally see the light of day.

Jason Kreher: Well, the business problem to solve was how to get this new tech into the hands of as many creative advertising people as possible. The challenge was that most generative AI video is garbage and everybody hates it. 

Our insight was that every creative has “the one that got away.” The script they never got to make because it was too expensive, the celebrity backed out, the focus group said it was weird, whatever. We figured if creatives could practice on something that never had a chance to be real, they’d instantly see all the ways they could use this tool in their process moving forward.

With around 400 entries to the competition, the brief has clearly inspired creators, were you surprised at the scale of the response?

CI: We were completely blown away by both the scale and diversity of the submissions. We saw entries from legendary industry veterans like Hal Curtis alongside brilliant work from creatives fresh out of college.

It’s no longer about who has the biggest budget; it’s about who has the best idea.

It highlights how the curiosity around AI within the industry is exponentially more pervasive than it was even 12 months ago. Creators are recognising that Luma is removing the traditional barriers to high-end production. It’s no longer about who has the biggest budget; it’s about who has the best idea. 

The quality of submissions was so exceptionally high that out of the 371 submissions, we ultimately selected over two dozen to submit to Cannes.

Above Real human actors were used – but not harmed – in the making of this film by Hal Curtis and Chuck McBride. Actors’ images were used with permission, licensed and paid for.

With AI tech offering unlimited realisation of concepts, have you noticed any particular trends or tropes emerging? How are artists responding to the freedoms it affords?

CI: When you look at submissions ranging from Bacteria Wars to Adrenaline Junkies, these are off-the-wall, one-of-a-kind concepts that previously would have required massive budgets, complex logistics, and months of production time. Now, they’re entirely possible. Artists are responding to this freedom by pushing the boundaries of what they pitch. AI is going to make the "impossible" idea the new standard.

JK: What I have noticed is that AI companies all show the concept of "anything is possible" with a rendering of a very realistic looking hologram tiger that is soaking wet. Basically just CGI porn. It's a very functional and limiting view of what you would create if you can create anything, and it shouldn’t be surprising that this turns people off. 

I remember David Droga saying that the camera was invented for scientists, but it only got interesting when the artists got hold of it.

I think we are only now starting to see how creative people are finding new ways to test the edges of what’s possible in this medium, and Luma has been the most unabashed in celebrating the people who make the work actually good. I remember David Droga saying that the camera was invented for scientists, but it only got interesting when the artists got hold of it. It's so early right now, but it’s going to get wild quickly.

Above: Director Ruth Bellotti envisaged Jesus running on water in one of the fictitious Luma ads submitted.

Video generation seems to have advanced well beyond the ‘uncanny valley’ in that the verisimilitude of ‘human’ beings has greatly advanced, will this herald in a new era of non-human actors?

CI: Not quite. We’re actually giving human actors more opportunities by utilising these advances to reinvent the production process. A great example of this is a new method of filmmaking, "Hybrid Production” as can be seen in our work with filmmaker-led production services/post/visual effects company Innovative Dreams. The first project to utilize Hybrid Production, shot entirely on a virtual stage, is The Old Stories: Moses, a three-part companion special from the world of House of David, starring Ben Kingsley.

Rather than replacing humans, Hybrid Production blends live-action human performance with real-time, AI-generated environments. Actors perform on a physical soundstage in front of an LED volume.

There’s something still kinda creepy about the eyes.

Filmmakers can instantly generate and project photorealistic environments, sets, and lighting onto the screens behind the actors while they are performing. This empowers directors to iterate on complex scenes instantly on set, slashes weeks off post-production timelines, and ultimately makes more ambitious live-action filmmaking financially viable.

JK: I am not sure it’s 100 per cent there yet with video tbh; there’s something still kinda creepy about the eyes, and it’s very hard to direct performance. But the leaps AI takes every month means this point will be moot pretty soon. And you’re already starting to see people get creative with how to make this tool work for the industry… the guys from Arts & Sciences made this spot (Adrenaline Junkies) with production company One Day by licensing the images of real human actors with their permission and paid them like a regular Screen Actors Guild job.

Again, this is very early days, but you can tell it’s absolutely going to change the game in ways we haven’t thought about yet.

Somehow many of the Dream Brief films have a strong sense of realness/authenticity. Is reality relative? Is it somehow in the story or the emotions of the piece, even if we know it is ‘fake’?

CI: It all comes back to the core idea. The vision and the intent behind these pieces are 100 per cent human. While the graphics and visuals are the product of highly sophisticated creative intelligence, the emotional direction that drives the production is human, and that is exactly what makes them resonate.

Above: Only weeks after his mother’s suicide, director Roman Jonsson heard about the Luma Dream Brief and created The Hug, hoping to win the Cannes prize money to fund the fight against depression.

CI: Our CEO, Amit Jain, often talks about how purely AI-generated video without a human perspective feels "dehydrated." It lacks humour, cultural relevance, and specific storytelling intent. What you see in films like Luma Pill or The Hug is "hydrated" content. The creators used AI as an execution tool, but the soul of the work — the emotion, the pacing, the authentic human truth at its centre – belongs entirely to the artist.

What have been the highs and lows of this project?

CI: High: The incredibly diverse range of content. Receiving 371 unique, fully-realised submissions that truly showcased the depth of storytelling our technology can enable was a massive validation of our mission. Low: It was agonizing to narrow down such an incredible pool of thought-provoking, one-of-a-kind pieces to the final 26 we selected to submit to Cannes. So much of the work was truly exceptional.

The reality of this tech terribly upsets a lot of business models, but I think it also makes for the most excitingly creative opportunities since the early days of the Internet.

JK: The lowest point was when our whole studio was filling out like three dozen separate Cannes portals and then verifying all of the information multiple times. But don’t tell them I told you that. 

The high point for me personally was talking to [Roman Jonsson] the creator of The Hug. It was such a raw, real story, and he said he really surprised himself by what came up as he worked on this. I knew we would get some creative interpretations of the brief, but for me this one really demonstrated how AI can help people realise an idea that was not possible otherwise. That's my favourite part about creativity… how it can just knock you over the head with something you never saw coming. 

Can you tell us something about the jury and the judging process?

CI: Our jury spans the best of advertising, production, and brand marketing (see the full jury here.) It ranges from creative legends and production company leaders to CMOs. The judging direction was simple: ideas first. The bar is originality of thinking.

JK: Our jury was stacked with a diverse set of creative legends across a variety of jobs and industries. And they were all loud opinion-havers, that was the unifying trait. We got Bill Oakley from The Simpsons, Katie Gurgainus from Nike, Richard Turley from Interview… Lora Schulson, Michael Hagos at HBO, Mal Ward, the legendary Carol Dunn at Barking Owl. Jeff Kling (who taught me how to write), Susan Hoffman who got me my job at Wieden. I could go on. 

All of them took this very seriously, and we got to a shortlist of 21 after a lot of heated discussion about the value of an idea, the importance of craft in execution, and what we were all actually judging.

Above: In the mock ad for the Luma Bar & Grill, diners are offered an obscene amount of multiplying 'Endless Shrimp'.

What type of qualities did your judges appreciate in the 21 selected films?

CI: The clearest pattern: a sharp, ownable idea tied directly to the product, executed with cinematic craft. The top-praised qualities were: Clear, ownable ideas; cinematic craft; sharp writing and storytelling;  humour that lands: emotional resonance; surprise and originality; ship-ready production quality; world-building that left judges wanting more; genre play as a Trojan horse, especially horror selling mundane products; hyper-specific details (casting, music, props) that pushed a film from good to great.

AI models are fundamentally prediction engines, but great comedy and poignant storytelling rely on the exact opposite — breaking predictability, shocking the audience and tapping into genuine human emotion.

JK: "The Big Idea" was the thing, no matter how clichéd it sounds. However. Now that the twin pillars of production and distribution have been democratised, an actual idea is the most important currency we have left. This was the topic that came up the most with our judges... the reality of this tech terribly upsets a lot of business models, but I think it also makes for the most excitingly creative opportunities since the early days of the Internet. It makes people who know how to spot a great idea more valuable than ever. 

What is it that separates the good AI stuff from the slop?

CI: The inherently human markers of any content – vision and intent – is and will continue to be what separates defining and groundbreaking work from “slop.”Purely AI-generated content is dehydrated –  it relies on just a simple text prompt and gets old very quickly. The sheer novelty of AI generation wears off fast, and what you’re left with is content that lacks a perspective.

The truly good AI content requires a creator's editorial control to inject humour, cultural relevance, and specific storytelling intent. Making something truly entertaining is incredibly difficult, and AI alone cannot replicate that nuanced human understanding.

AI is fast as fuck,  and it can easily mimic the patterns of humour.

JK: I’ll paraphrase US Supreme Court justice Potter Stewart: “I shall not today attempt further to define the kinds of material I understand to be embraced within that shorthand description, and perhaps I could never succeed in intelligibly doing so. But I know it when I see it.”