I Rebuilt Tehran With Google Maps and AI. Here's Why. - Google Docs

She’s a deaf DJ playing illegal techno sets in underground Tehran. I’ve never set foot in the country. This is how I made a film about it anyway.

A year ago, I almost skipped that panel. It was the Cannes Film Market, and I’d stumbled into a room of international filmmakers for no particular reason. One of those unplanned detours you take when the schedule falls apart. I ended up sitting next to Faeze : a young Iranian director who’d somehow made it to the South of France to show her work.

We talked. And at some point, the conversation stopped being about cinema. I asked her what life was actually like in Tehran. Not the political version. The real one. What she described was a city living in two layers. On the surface: restrictions, surveillance, conformity. Underneath: something else entirely. Secret clubs, underground movie theaters, entire social lives conducted in the dark. Young Iranians weren’t just bending the rules, they were building a parallel culture beneath the one the state could see. It was dangerous. Arrests happened. And they went anyway.

I didn’t know what to do with that conversation. I tucked it away.

Then the war erupted in February, and it came flooding back.

The Film That Needed to Exist Right Now

One of the things nobody tells you about AI is this: it’s not just a tool. It’s a time machine.

Traditional filmmaking runs on 18-month production cycles, minimum. By the time a film about a political crisis reaches an audience, the crisis has moved on. The world has forgotten. But AI collapses that timeline. A filmmaker with the right tools can respond to the present while it’s still present.

That's why Foujan: The Sound of Silence exists.

After joining the Flick Film Residency, I wanted to make something that resonated with the reality of a generation living underground, communicating without words, dancing in secret. The film became my answer to Faeze’s story.

What follows are the decisions. aesthetic, technical, human, that shaped it :

The Character Who Unlocked Everything

Her name is Foujan. She’s a DJ. She’s hearing-impaired. And she plays techno sets in an illegal underground club in Tehran, feeling every beat through the vibration of the floor, the walls, the speakers pressed against her skin.

When I landed on that character, everything else in the film snapped into place. Because here was the thing: the story I wanted to tell was about a generation that communicates without words, that lives in frequencies the state can’t detect, that finds freedom in sound while officially living in silence. A hearing-impaired DJ wasn’t just an interesting character. She was the controlling metaphor for the entire film. She experiences the world exactly the way that world works, not through language, but through vibration. Not through what is said, but through what is felt.

That’s the film. No dialogue. No explanation. Just vibration, then rupture, then silence.

Silence as a Language

I have no music training. None. Every piece of the Foujan score — the techno builds, the earth-frequency drones underneath them, the moment where both collapse into silence — was generated with AI.

And for months, I genuinely didn’t know if it would be enough. A film with no dialogue lives or dies by its sound.

The doubt was real. There were weeks where I was certain the whole thing was a mistake. What kept me going was the discovery that techno music and low-frequency earth vibration follow the same rhythm. The same intervals. The same pulse. That’s not a metaphor I invented, it’s physics. And once I understood that, the score stopped being a collection of generated tracks and became a single continuous argument: that the underground clubs of Tehran aren’t an escape from the world.

Foujan doesn’t hear the music. She is the music. That idea was worth the doubt.

Building a World That Feels Like a World

I generated over a thousand images before I committed to a visual style.

That number sounds obsessive, and it probably was. But this kind of obsession matters. Every image is a question: Does this feel true? Most don’t.

The final choice : a deliberately flat, graphic visual language, was a rejection of the obvious. Hyperrealism was available. AI can produce photorealistic frames that are nearly indistinguishable from live footage. But photorealism would have been the wrong answer. The film isn’t trying to document Iran. It’s trying to interpret it. The stylization is a statement: this is not the world as it is, but the world as these characters feel it.

Flatness, it turns out, can carry tremendous emotional weight. The less visual noise there is, the more the sound fills the space.

Tehran, Reconstructed

Here’s the problem nobody warns you about when you set out to make a film set in Iran: you can’t film there.

It is illegal. It is dangerous. Filmmakers have been arrested for less. So the question becomes: how do you honor a real place without ever being allowed to stand in it?

My answer was Google Maps. I used street-level footage of actual Tehran locations, the entrance of the Grand Bazaar, a Persian park in the city center, and ran it through AI tools to transpose it into the film’s visual style. Real geography. Real architecture. Real streets. Transformed, not invented.

This felt important. I could have built a fictional Middle Eastern city from scratch. It would have been easier. But it also would have been a lie, a comfortable, imaginary version of a place that real people are navigating right now, in ways that put them at genuine risk. Using the actual streets of Tehran, even refracted through AI, felt like the only honest choice.

The Hardest Part: Getting the Voice Right

I don’t speak Persian.

That’s an uncomfortable thing to admit when you’re making a film set in Iran, with Iranian characters, in the Persian language. And no AI tool in the world will save you from that gap if you haven’t done the work to close it yourself.

Before a single line of dialogue was generated, I watched all three seasons of Tehran — the Apple TV series — specifically to train my ear. Not for plot. For sound. For the cadence of Persian, the way emotion shifts the rhythm, the texture of the language when someone is scared versus when they’re defiant. Only after that immersion did I feel ready to make decisions in ElevenLabs about which voices, which intonations, which generations were true and which ones were convincing imitations of something I didn’t actually understand.

This is the part of the AI conversation that tends to get lost. People argue about whether AI makes us more or less human. I think the question is backwards. The tool doesn’t determine the answer. The filmmaker does. Al gave me the ability to make a film in Persian. My responsibility was to earn that ability before I used it.

Imitate, then innovate — as someone once said.

What Foujan Is Really About

I made this film because of a conversation I almost didn’t have, in a room I almost didn’t enter, with a filmmaker I almost didn’t meet.

Faeze didn’t set out to teach me anything that afternoon in Cannes. She was just telling me about her life. But what she described : a generation that refuses to be silent, even when silence is what’s demanded of them, became the spine of everything.

Foujan is not a political film. It doesn’t argue. It doesn’t explain. It pulses. And if it works, it’s because the technology was in service of something human : a story that needed to be told, from a place I couldn’t visit, in a language I had to learn to hear.

Here’s what I took from making it: Al doesn’t flatten culture. But it will expose you if you let it. The tools are fast. The responsibility to use them carefully is not.