From borrowing cameras and struggling to finish a short film in 2012 to creating a cinematic sci-fi story with AI and my kids in 2026. What a change in technology!

Sometimes, you need to try something yourself to truly understand how much technology has changed.

I’ve been experimenting with AI for quite some time, from generating images and training personal models to exploring AI-assisted engineering and creative workflows. But recently, I tried something that made me stop and think about just how far we’ve come.

I created a two-minute science-fiction short film featuring my own avatar, a self-built spaceship, an AI companion, a little comedy, and a mysterious transmission from the far side of the Moon.

The time I spent?

Around 30 minutes developing the story and another 30 minutes generating the video. In total, roughly 1 to 1.5 hours.

And I did it with my kids.

To understand why this felt so incredible, let me take you back a little.

2012: When Making a Short Film Was a Different Kind of Challenge

I still remember trying to create a five-minute short film back in 2012.

It was an exciting idea. I had a story in mind and wanted to turn it into something real. But turning that idea into a finished film was an entirely different challenge.

I spent nearly a month working on it.

I had to borrow cameras, coordinate with actors, request that everyone arrive on time, manage the shooting schedule, and work around the practical limitations of the equipment and people available.

Even after putting in all that effort, things didn’t always go as planned.

Recorded clips went missing. Schedules fell apart. Getting everyone together at the right time was difficult. Small problems kept adding up, and the final result never quite reached what I had originally imagined.

Eventually, I had to accept that I couldn’t finish the film the way I wanted.

The experience taught me something about filmmaking: having an idea is one thing; bringing it to life requires resources, coordination, time, and a lot of patience.

Back then, even a simple five-minute film could become a month-long project.

I never imagined how different that experience would become just a few years later.

2024–2025: Generating Images of Myself With AI

Fast-forward to the time when AI image generation started becoming a serious creative tool.

I still remember experimenting with Flux image models and using them as base models to generate images of myself.

The idea was fascinating. Instead of relying on generic AI-generated faces, I wanted to create images that actually looked like me.

But getting consistent results required more than simply typing a prompt.

I experimented with my own photographs, prepared datasets, worked on captions, trained a LoRA model, and tested the results using tools such as ComfyUI.

It was a technical process as much as a creative one.

In April 2025, I published my work on ArunUK V2, my personal AI face model trained using Flux-Dev as the base model.

I documented the process, shared the model, and wrote about the experience on my blog.

You can read that post here:

Next-Level AI Face Model: ArunUK V2 Is Now Live

I used around 30 photographs of myself, prepared detailed captions, trained the model, and experimented with generating realistic images.

Looking back, that was already a significant step forward.

I could create different versions of myself in different settings, explore visual ideas, and experiment with creative possibilities that would previously have required a photographer, a location, lighting equipment, and considerable editing.

But there was still a distinction between generating an image and creating a film.

An image was a moment frozen in time.

A film needed movement, continuity, characters, scenes, and a story that held together.

And now, that gap is getting smaller.

2026: From a Photograph to a Science-Fiction Film

Recently, I decided to experiment with Google’s AI video-generation capabilities using my avatar.

I didn’t start with an elaborate production plan. I didn’t spend days writing a screenplay or preparing storyboards.

I started with a simple idea.

What if I were a lone scavenger surviving between Earth and the Moon in the year 2189?

Humanity has colonized other planets. Abandoned spacecraft and satellites float through the darkness. My character survives by collecting whatever useful equipment he can find.

He has a small spaceship that he built himself.

And the only companion he has is an AI named MIRA.

She’s sarcastic, intelligent, and occasionally annoying. She helps keep the ship running, monitors oxygen levels, and complains about my questionable survival decisions.

Together, we navigate the dangers of space.

There’s a little comedy, a little affection, a dangerous salvage mission, and a mysterious transmission that might change everything.

That became THE LAST SCAVENGER, my two-minute science-fiction short film.

What surprised me wasn’t just the ability to generate the video.

It was how quickly I could move from an idea to something I could actually watch.

I developed the story, divided it into short scenes, prepared the video prompts, and generated the clips.

My kids joined me in the process, making the entire experience even more enjoyable.

There was no need to coordinate actors, borrow cameras, arrange locations, or organize a physical shoot.

Instead of spending most of my time managing production logistics, I could focus on the creative decisions.

What should happen next?

What should the character say?

How should the AI respond?

What would make the ending memorable?

And just like that, I had a short film.

The entire experiment took roughly an hour to an hour and a half.

Compare that with my experience in 2012, when I spent nearly a month trying to create a five-minute film and still couldn’t finish it as intended.

That’s the difference that really struck me.

Watch the Experiment: THE LAST SCAVENGER

I’ve embedded the resulting short film below.

In a universe full of people, loneliness is the one thing you can’t escape.

The story follows a lone scavenger trying to survive with limited oxygen, a patched-together spaceship, and an AI companion who might be the closest thing he has to family.

It’s a small experiment, but it represents something much bigger to me.

Watch THE LAST SCAVENGER — My AI-Generated Sci-Fi Short Film

I know I could have made it more polished.

I could have spent more time refining the script, improving scene transitions, adjusting the dialogue, experimenting with camera movements, and generating better shots.

I could have iterated on the visuals and worked harder to make every scene consistent.

But I deliberately kept this experiment small.

I wanted to see what was possible without turning it into another month-long project.

And I wanted to enjoy the process with my kids.

For me, that was enough.

The Biggest Change Isn’t Just the Technology

When people discuss AI-generated images and videos, the conversation often revolves around the quality of the output.

How realistic does it look?

How good are the movements?

Can it produce cinematic lighting?

Can it generate convincing dialogue?

Those are important questions, but I think there’s something even more significant happening.

The cost of experimenting with an idea is falling dramatically.

Think about what was required to make a short film in 2012.

You needed equipment, actors, coordination, locations, time, and the ability to bring everything together.

Now, one person can develop a concept, generate visual scenes, experiment with different approaches, and produce something that can be shared with the world.

The same principle applies beyond filmmaking.

Someone can explore a product idea without immediately building the entire product. A developer can test a concept before investing weeks in implementation. A teacher can create visual explanations that would previously have required specialized resources.

The possibilities are expanding because the distance between imagining something and attempting to create it is becoming shorter.

That doesn’t mean every result will be good.

It doesn’t mean storytelling, filmmaking, engineering, or artistic judgment have become irrelevant.

Quite the opposite.

When the technical barriers become lower, the quality of your ideas and the decisions you make during the creative process become even more important.

AI can help generate a scene, but deciding why that scene belongs in the story is still a creative decision.

AI can help create a character, but making the audience care about that character requires more than visual realism.

AI can generate a film, but whether that film is worth watching depends on what you do with the technology.

From Learning the Tools to Exploring the Possibilities

My journey with AI has changed considerably over the last couple of years.

I went from experimenting with Flux models and training my own face model to generating videos featuring my avatar.

The underlying tools, interfaces, and workflows continue to evolve.

What once required specialized knowledge, experimentation, and considerable setup is becoming accessible through increasingly simple interfaces.

Today, I can upload an image, develop a story, and begin experimenting with an entirely fictional world.

Tomorrow, the process will probably become even more capable and convenient.

But I don’t think the most interesting question is how much faster AI can generate something.

It’s what people will choose to create when the barriers are no longer as intimidating.

How many ideas never became films because someone couldn’t afford a camera?

How many stories remained unwritten because the creator couldn’t assemble a team?

How many people gave up on creative projects because the technical work was too difficult?

We may never know the answer.

But we’re entering a time when more people can at least attempt to bring those ideas to life.

And that’s exciting.

A Small Experiment, With a Big Lesson

I didn’t spend weeks on THE LAST SCAVENGER.

I didn’t aim to produce a perfect film.

I simply wanted to experiment, explore the technology, tell a little story, and have fun creating something with my kids.

In about an hour, I went from an idea to a two-minute sci-fi film that I could share.

I still have plenty to learn about AI video generation. There are obvious areas where I could improve the result, and I can already think of several ways to make the next experiment better.

But that’s precisely the point.

I can try again.

I can experiment with a different story, a new visual style, a more ambitious idea, or a completely different genre without committing an entire month to the process.

And the next time, perhaps I’ll create something even better.

For now, I’m happy with this little experiment.

Because sometimes, the most interesting thing about new technology isn’t what it can produce.

It’s what it finally allows you to try.

In 2012, I spent a month trying to make a five-minute film. In 2026, I made a two-minute sci-fi film in roughly an hour, with AI and my kids.

What a change in technology.

And I suspect we’re still only getting started.

---

Have you experimented with AI-generated images, videos, music, or other creative projects? I’d love to hear what you’ve been creating and how your workflow has changed.

What would you create if you could turn an idea into a film in an afternoon?