AI and Filmmaking — What’s Really Changing?

I wanted to write about AI for a long time but thought I should explore more before doing that. Yesterday, I saw a music video made entirely with AI, and that finally pushed me to write this piece.

I’ve been working and experimenting with AI for quite some time now. Some of you who know me might have already seen some of my work. I started with Text-to-Image and Image-to-Video tools just for fun, more like an experiment. Those were the days (not during the dinosaur era, of course, but just a few months back) when AI used to generate weird stuff (people with extra limbs, distorted faces, and all that.)

The tool I was using then was GenCraft, a basic text-to-image AI. It wasn’t perfect, but the quality was decent enough for my presentations and shot descriptions. As a filmmaker, I was genuinely excited.

I started in the industry when Hi8 was fading out and Beta was taking over television. For films, Arri cameras were used with Kodak film stock. Editing was done on Avid, which was a huge deal back then and an expensive machine, costing around ₹55 lakhs or more depending on the setup. 

Making films or TV projects wasn’t easy. You had to convince so many people, wait for approvals, and raise money, even for a short film. I’ve seen people spend years just trying to get someone to listen to their script.

When I was in film school, we were taught on digital cameras, the Panasonic HVX200 with P2 cards was very popular because it was affordable. Later, we worked on Panavision Gold II, a proper 35mm film camera. Shooting on film was an amazing experience. Sitting behind that camera felt like holding a cannon; it gave you a special kind of thrill.

We also worked regularly on Arri 16mm cameras, threading magazines became second nature. When it was time to choose between shooting on film or digital, I chose film, because it was a process we loved. Choosing stock, shooting, waiting for it to be developed, that was exciting.

The ones who chose P2 could shoot in the day and edit by evening. We who chose film laughed at them because they were missing the process.

When I came back to India, I met an old friend, a senior and an award-winning cameraman. I told him how digital cameras were getting cheaper and would soon change everything. He smirked and said, “A film not shot on film is not a film, it’s a video.”

He gave me a lecture on how digital would never take over. He was a big name, so I doubted myself and thought maybe I was expecting too much from digital.

A few years later, when digital completely took over, I met the same guy again, now as a co-producer on a film. He had already ordered several hard drives but also bought Kodak film cans. I asked, “We’re shooting on digital, so why the cans?” He said, “We’ll take some outdoor shots on film, digital can’t handle that.”

We shot for two months outside Mumbai. Not a single film can was opened. After the film, he चिपकाओड  those cans to someone else with the same story.

Kodak nearly collapsed after that. We all know how that story went.

Everything changed with digital. Hard drives replaced magazines. Filmmaking didn’t become easy, but it definitely became faster and more accessible.

When AI came, people didn’t take it seriously at first. But once text-to-image and image-to-video tools appeared, things started getting interesting. Then came Google Veo models, and suddenly everyone was talking.

Soon, everyone started making random, meaningless videos. Low-quality, rushed content, just made for clicks. That’s when many people got put off. “AI means nonsense content.” That became the mindset.

On the other hand, the kind of cameramen I spoke about earlier, the traditionalists, started saying AI would “kill filmmaking.”

It won’t. It will uplift filmmaking, but only if we know how to use it.

Ask yourself, who’s making all this bad AI content that’s giving it a bad name? Are they real filmmakers, or just random people chasing quick money and views?

A serious filmmaker will use AI wisely, as a tool, not as a shortcut.

Right now, AI is still in its early stages. Most tools are offering limited free access so they can train their models better. Every random video we generate helps them improve. We’re part of that training process, knowingly or unknowingly.

Somebody sent me a Facebook post where a person wrote, “Let’s hope this AI bubble bursts faster so that we can make real films.”
That made me think — are we making the same mistake again? Are we taking AI too casually?

It’s not going anywhere. Once the training phase slows down and the tools mature, prices will rise, the quality will improve, and all the useless stuff will fade away. We’ll have no choice but to adapt.

There are already so many filmmakers doing incredible work with AI that would blow your mind. You probably haven’t seen them because they aren’t viral yet, but they’re out there.

What actually made me write this was Diljit Dosanjh’s new song “Sorry,” uploaded by T-Series. The whole video is made with AI.

As someone who supports generative AI, I should have been happy. But honestly, I wasn’t. The video looked mediocre. I don’t know how a big label like T-Series approved it. AI has gone far beyond what that video showed.

It has around 150k views, and I read the comments. One said, “Ha ha ha, AI se banaya hai.”
That’s the mindset right now, as if using AI is “cheating.” “Pakad liya”

T-Series never claimed it wasn’t AI, but people still saw it that way. They think you just type one line, press a button, and boom, a film pops out. That’s not how it works.

Yes, some people do that, but that’s not filmmaking.

You need to tell a story. The medium doesn’t matter, film, digital, or AI.

We once laughed at digital too. Now it’s the backbone of cinema. AI will go through the same ridicule, misuse, and fear. But eventually, it’ll become just another tool in a filmmaker’s kit.

What matters, as always, is not the camera or the code. It’s the person holding it.

- Viveck Tewari