Missed Connections: On Making a Film With AI

This is a story I’ve been carrying for a long time — probably 20 years or more. It always lived in my head as something that would happen in Paris, where I’m from. Paris is a place that naturally lends itself to romantic storytelling — whether through photography, film, or just the way people move through the city.

The subway, in particular, has always felt to me like a perfect metaphor for life. You make a choice — take one train, not the other — and that choice changes everything. People pass by, sometimes just for a second, and then they’re gone. Maybe forever. That idea of a fleeting encounter, of something that could have been but wasn’t — that’s stayed with me. And I think it stays with a lot of people.

The movie Sliding Doors illustrated this eloquently. So this film became a way for me to capture that feeling. The missed moment. The wave across the tracks. The quiet heartbreak of life ruthlessly moving forward.

Inspiration: Paris and New York

It’s a mix of Paris and New York — two cities that shaped me both personally and creatively. I discovered New York at 14 and kept coming back. It became the city that inspired me most, especially cinematically. But Paris is home, and it’s where my sensitivity to urban moments was first shaped.

I’ve been a photographer for over 30 years, and my style is very specific: I shoot people from behind, often in motion, always in urban settings. No faces. The point is to leave space — so the viewer has to bring something of themselves to the image. It becomes a conversation rather than a statement.

That approach naturally fed into how I started thinking about film. The mystery, the framing, the tension of not saying everything — that’s what draws me in. And the subway, especially, is a place that contains so many of these layers. It’s a space where everyone crosses paths — briefly, anonymously. Even when it’s not romantic, there’s a spark of curiosity: Who is this person? What’s their story? That question is where empathy and imagination begin. And for me, that’s also where storytelling begins.

Why I Chose Runway

I started experimenting with a bunch of AI tools that were great at generating one-off images, but terrible at keeping character consistency. And if you’re trying to make a short film — even just three minutes — that’s essential.

I found that Runway was one of the best-rated tools for video generation with character consistency. That’s what got me into it.

But I came into this with no background in filmmaking — so I had to learn the tools, and also teach myself a whole new language. I had to figure out how to speak in cinematic terms: what’s a dolly-in versus a zoom? What’s an extreme wide or a bird’s-eye view? Those are storytelling tools. And I had to learn them from scratch. I am still very much a beginner.

What made Runway work for me was the ability to generate an image, then animate it, then iterate. It let me build shots that connected, evolve them scene by scene, and slowly piece together the film I had in my head — even if I was learning while doing.

The Process, Step by Step

The story was always clear to me. I had this scene in my head — two people on opposite platforms, one train arrives, one leaves, and they never see each other again. But I wanted there to be a twist. A second chance.

The part that wasn’t clear was: how do I actually tell it? I could see the first and last shots in my mind, but filling in the rest — that’s where the real work began.

I didn’t want the film to feel like it was just AI-generated. I wanted it to be grounded in reality — in places I know and love. So I approached it as a hybrid process. I’d take real photos of metro stations that inspired me — places I’ve walked through a thousand times — and then build the scenes using those as anchors. I used my own photography as the foundation, and then used Runway to bring characters into those environments and animate them.

In the end, the first iteration of this film has about 20–30 shots in the Montreal subway. Some are fully AI-generated, but many are built around real-world frames that mattered to me. That’s how I stayed connected to the story — and made it feel real, even while using new tools.

The Hardest Parts

There were a few, honestly.

The first was technical — learning how to use tools I’d never touched before. Image generation, video generation, editing — all of it was new. But because I was genuinely curious and excited to learn, it felt more like play than work. If I’d had to learn accounting software, it would’ve been brutal. But this? This was fun. Still hard, but fun.

The second challenge was learning the language. Not English or French — but the vocabulary of film. I had to teach myself how to “speak camera,” how to direct movement, and how to guide a character’s behavior through prompts. With AI, it’s not enough to have an image in your mind — you need the right terms to get the model to do what you want.

And then there was the self-doubt. This was my first film. I’d never written a script and turned it into something visual before. There were moments where I’d compare my work to slick, cinematic sci-fi clips someone else posted and think, Why am I doing this? Who’s going to care about my little three-minute story?

But I’ve learned that self-doubt is part of the path — especially in creative work. You just have to stay focused on expressing what you want to say, how you want to say it. Some people will connect with it. Some won’t. But if it’s honest, and you’ve crafted it with care, that’s what matters.

Did AI Change My Creative Voice?

I’m not entirely sure yet — but it definitely expanded it.

For years, I’ve been telling stories through photography. A single shot. One frame to suggest a mood, an emotion, a question. I’ve always loved that constraint. But with film — especially through AI — I realized I could finally go beyond the suggestion. I could actually tell the story.

With AI, I got to follow through — to create the next frame, and the next one, and let the story unfold. That was new. And honestly, it felt like something I’ve been waiting to do for a long time.

What’s funny is that as a street photographer, you don’t get a second take. It’s point-and-shoot. But with AI, I suddenly had thousands of generations — maybe 1,000 or more across images and video. I could iterate. Direct. Adjust. That changed everything.

What People Don’t Realize About AI Films

It’s much easier than you think — and much harder than you think.

It’s easy to generate something that looks good. But what’s hard is getting the model to do exactly what you want. That takes precision, clarity, and a kind of linguistic discipline.

Consistency is another challenge. Tools like Runway help, but maintaining coherence across shots — especially from scene to scene — requires planning and prompting skill.

And then, there’s the solitude. You’re working entirely alone. That comes with freedom, but also with the need for a kind of inner balance — you have to be your own critic, your own producer, your own motivator.

What’s Next

This first one is set in Montreal. But I’ve always imagined telling it in Paris, in New York. Now I want to expand it: Tokyo, Moscow, maybe more. Each city has its own emotional palette — its own tempo, its own way of holding space between strangers.

The goal is resonance. To show how something as small as a glance or a wave can mean something completely different depending on where you are.

And with AI tools evolving so quickly, the next version will be faster to make, richer in detail, and more nuanced in performance.

🎬 Watch the film here: “Missed Connections”