My AI Movie Breakthrough

I have long dreamed of making an Imperial Norway movie. But I’ve worried it wasn’t technically possible—that it would be too difficult.

My latest Tucker-James Goes to Imperial Norway video has convinced me that I actually can do this. Not in the style of a Hollywood movie—I don’t believe current technology will ever replace filmmaking with real actors on real sets. Some optimists believe that, but I’ll explain why I think that will simply never be possible.

Instead, this will be something different—movies made within the opportunities and limitations that AI offers. It’s not about replicating Hollywood, but about creating something compelling that allows me to tell an engaging story you can follow.

AI has many issues, such as instability in images and characters. Faces change. Even body shapes shift throughout a video. When a person walks back and forth in a room, the parts of the background that were covered and then revealed will often look different.

Because of this, I think AI as a medium must be thought of more like a dream. A world that constantly shifts. My first breakthroughs in AI imagery came from accepting its limitations—and learning to work with and around them.

You have to think differently when telling a story through AI than with other media. Since faces can change, I identify characters by their voice, hair color, hairstyle, and clothing. That’s how viewers can still recognize who’s who.

All media have limitations. A book can’t show you images or smells, but authors find ways around that. Even traditional movies have limitations. I remember someone once told me movies are better than books because they have visuals and sound. They believed movies could do everything a book could—but they can’t.

Movies are about the external: what people say and do. They don’t give you internal thoughts directly like a book. So filmmakers have to approach things differently. They weave thoughts and feelings into dialogue and visual cues.

I remember this from the Game of Thrones series. People often say that movies should just adapt books faithfully, but that’s naive. It’s an entirely different medium. Dialogue in Game of Thrones was often changed to express feelings and thoughts that the book simply stated outright.

I had to remind myself of this while making my latest short film. There was a clip where Pastor Elbert R. Strickland said, “Hey, son, how’s your trip to Norway going?” I replaced it with, “Hey Tucker-James, how’s your trip to Norway going?”

Why the change? In a book, the first version would work fine because you can add something like: Elbert said to his son, Tucker-James… Books can give you information outside the characters’ dialogue. In a movie, much of that has to come through the dialogue itself.

This was my way of telling the viewer who this young man is. I could’ve used a text label on screen, but that would kill the sense of immersion. This is what’s meant by “show, don’t tell.” Avoid info dumps; let the audience figure things out naturally.

Instead of a narrator saying, “This is Tucker-James,” you learn who he is from the conversation he has with his dad.

I’m using this example to highlight that every storytelling medium has different limitations and ways of working around them. A movie, novel, comic book, poem, music video, or song—they each have their own strengths and weaknesses.

We must think of AI video as its own storytelling medium, not as a substitute for traditional films. Think of how a poem and a novel are both text, yet written completely differently. They follow different rules and practices.

Many people struggle with AI art because they try to replicate other forms. They can’t accept the limitations and fail to work within them. Let me give an example. When Tucker-James sits in a cab, any reflection shows he couldn’t possibly be sitting inside the same car shown from the outside. The window layout doesn’t match.

A purist trying to perfectly replicate traditional movies would spend hours trying to match the interior and exterior. If you do that, you’ll find that AI saves you no time—you’d probably be better off learning 3D modeling or using green screens and live footage.

But most viewers will go along with the idea that Tucker entered this cab because the sequence of clips suggests it. The interior is similar enough to the car shown from the outside. It sells the illusion.

Likewise, at the beginning of the short film, the Kurebakken sisters talk to each other—but clearly, if you think about it, they’re not in the same kitchen. The backgrounds don’t match.

Still, most viewers can suspend their disbelief and accept they are. The dialogue and editing suggest they’re in the same place. As long as the visual style is similar enough, it works. If one were in the kitchen and another outside in the backyard, that would feel wrong. But if it’s “close enough,” people accept the vision.

I want you to think of it less as watching a movie and more like watching a dream. In dreams, we misremember details. The world is vague and shifting. You remember that someone was in a kitchen talking, but not exactly what that kitchen looked like.

Rather than spend hours matching every detail of a background, it’s better to accept this limitation and focus on conveying the idea that they’re in the same space—through clips and dialogue.

Some of you might say, “It’s just a matter of time before technology solves this.”

Sorry to disappoint, but that’s a long way off. Maybe someday, but not soon. What makes modern AI so powerful today is also what causes its flaws. When you describe an image, there are always missing details. The AI must imagine those details: the shape of a belt buckle, the way light falls through a window, the exact eye shape.

You might think, “Just give it a longer description.” That won’t help. It’s a fundamental limitation of how AI is trained—and of natural language itself. AI is trained on images and brief descriptions, which are never fully accurate. If a description could encode all visual detail, it would mean we’ve invented a magical image compression method—reducing an entire picture to a few words.

But that’s impossible. Language is inherently imprecise. Even if you hire a director to make a movie, it will never be exactly what you imagined. And what he makes won’t even match his own vision exactly—it depends on how the actors interpret his instructions.

Great actors bring something of their own. They don’t just follow orders like robots—they interpret, create, and contribute artistically.

That’s why I accept the medium I work in. I’m not competing with movies—any more than books compete with comics or films. AI video is its own form of artistic expression. What matters is whether I can use it to express a story—to give a sense of a world inhabited by characters with real personalities, emotions, and motives.

And that’s the breakthrough I’ve had. I’ve now convinced myself—proved to myself—that I know how to tell stories through the medium of AI video.

My goal isn’t to compete directly with traditional filmmakers. I want to make something enjoyable that leaves you with a sense of story and character. I want Imperial Norway to feel like a real place in your mind—even if it’s all imagination.