I don't decide film by film. I decide shot by shot. The question is never "Can AI do this?" These days it can do a surprising amount. The question is what this particular moment needs in order to feel true, and which tool gets me there with the fewest compromises. Some moments need a real face in real light. Some need a world that doesn't exist. More and more of the films I want to make need both.

People ask me which side I'm on. I'm a director. I'm on the side of the film.

The question I ask before every shot

Before I think about tools, I write down what has to be true in the frame. Usually it's one of four things:

• A performance. A look that lasts half a second too long. A breath before a line.

• An object. The product, exactly as it is: its shape, its material, its logo.

• Physics. Water, fabric, smoke, food, skin. Things the eye has watched its whole life and knows by heart.

• A world. A place, a scale, a time of day or a time in history.

Once I know which of these the shot lives on, the choice of tool is mostly made for me.

When I shoot it on set

If the shot lives on a human performance, I want a real actor in front of a real lens. Generated faces have become very good at looking like people. They are still not very good at being someone. The small things are where an audience decides whether to believe you: a hesitation, a smile that arrives late, eyes that change before the mouth does.

The same goes for products. When a brand needs its product shown exactly as it is, I shoot it. A bottle that subtly changes shape between two cuts, or a logo that melts for a frame, isn't a stylistic choice. It's a mistake the brand will see before anyone else does.

And physics. Pouring, splashing, wind in a sleeve. You can make these with AI, and sometimes beautifully. But when the whole shot depends on the eye believing the weight of something, a real camera on a real event still wins.

When I make it with AI

If the shot lives on a world, AI opens doors that used to be closed by budget, weather or physics. A city that doesn't exist yet. A desert at an hour the sun never reaches. A crowd of a thousand, a camera move no crane could make, a decade we can't travel back to.

I also use AI long before the final film. It lets me test an idea in images while it is still cheap to change, and show a brand how a scene will feel before anyone books a location. That alone changes the quality of the conversation. We argue about the film, not about a mood board.

What I don't do is use AI because it's faster. If a shot is easier to make with AI but worse, it's worse.

When I combine them

The most interesting work happens in between. A real actor shot on set, standing in a world that was built afterwards. Wide shots made with AI, cut against close-ups filmed with a lens and a person. A scene designed and timed with AI images, then shot for real because now we know exactly what we need.

When it works, nobody watching can tell where one tool ends and the other begins. That's the goal. Not "an AI film." Not "a traditional film." A film.

The rules that don't change

Whatever the tool, a few things are not negotiable for me.

Continuity. If a character's face changes between two shots, if a room rearranges itself, if the light jumps from morning to evening inside one moment, the film is broken. With AI this is the hardest part, and it's exactly where most AI work falls apart. I treat continuity of character, location and light as the first job, not the last fix.

It must not look like AI. If the audience notices the tool, I've failed. The standard I hold the work to is the one I'd hold any film to: does it look shot, lit and directed by someone who cared?

The idea comes first. A strong idea, researched properly, survives any tool. A weak idea doesn't get stronger because it was expensive or because it was generated.

Sound carries the story. People remember how a film felt, and half of that feeling comes through their ears. Music, silence, the room tone under a line. I give sound the same attention as the image, whichever way the image was made.

What I tell brands in the first conversation

You don't have to choose between "a shoot" and "an AI film." Choose the film you want people to feel. Then let me choose the tools, shot by shot, and I'll tell you honestly why. That includes the times AI is the wrong answer, and the times it saves you a shoot you don't need.

AI is a camera, not a shortcut. It's one more way to point at the thing that matters. The thing that matters hasn't changed.

Sabit Başman is a director and the founder of Chamber Eight, an independent, director-led creative company in Istanbul. If you have a film in mind, write to hello@chambereight.com.