AI artist Gokul Pillai shares how artificial intelligence is transforming photography

An AI-generated image of how Lord Ram must have looked at 21 has gone viral on Twitter. Chiseled features, dewy, almond-shaped eyes, lustrous hair, AI’s imagined version of the much-revered Hindu deity is so ridiculously handsome, it can induce an identity crisis even in the most strapping and successful models.

Another set of images of Donald Trump being arrested by cops and Francis Pope wearing a white puffer jacket has got wildly popular too. They were so on-point that the Pope was forced to clarify they were generated using artificial intelligence.

Most recently, yet another series of images, that of the world’s richest men imagined as living in slums, sent social media into a frenzy. The series features several celebrity billionaires, including Trump, Bill Gates, Mark Zuckerberg, and Jeff Bezos, covered in grime and surrounded by poverty, a scathing critique of the ever-widening gap between those with and without wealth.

The man behind these images is Gokul Pillai, an IT consultant living in Mysore with a massive following on Instagram, courtesy his finesse with AI and photography. His IG account is a curious amalgamation of AI art and old-school photography. This World Art Day, the 32-year-old discusses how AI can create disruptive art and alternate realities. He also talks about his craft, process, AI being misused to spread fake news and misinformation, the threats of deepfake, and how AI will shape the future.

I have been doing photography and working on digital art for over eight years now. So I already had a base. I have been posting on Instagram since 2015. So I was always there in this field but last year, when Midjourney was introduced, there was a lot of hype around it. I wanted to try it out too but the initial results were not that great. So I left it. Later when the newer versions came, the pictures they produced were very realistic. So in November last year, I started putting in more time.

No, no. I just learned from YouTube tutorials and whatever free resources were available online.

I generate the initial image from Midjourney. If I want to make any changes, I use Photoshop. I do some color grading as well on all my images. That I do in Lightroom.

Yea yea, extremely. The pictures that you directly get from AI apps are not aesthetically pleasing, at least not to my liking. My photos always follow a particular style. I edit my pictures in a certain way. So I do the same thing with the images that I generate using AI also so they have coherence with the other things that I post.

So for instance, I created a picture of an old man holding a small toy car. For it, you can write a middle-aged man holding a toy car. The surroundings should look like an old garage. There should be a window on the right. The light should come from this particular direction. His eyes should be closed. You need to describe what you want to see in the picture as much as you can. The more specific you get, the better the results. I usually give a basic prompt first. Based on the result, I tweak it. You can even upload the initial image that was generated back into the program and change certain areas.

It really depends on the scene. But if I want to do something very specific, I need to redo it about eight or nine times or even more than that in certain cases. You hardly get anything on the first go. And then there are times when you might get something that you hadn’t imagined but you like it more than what you had initially thought of. So you can use that as the starting point and take it from there. I usually have a story in mind that I narrate through a series of six-ten pictures. It takes about three days to create an entire series. I usually work at night in two-three hour slots.

One day, I saw a post on the internet of superheroes imagined in everyday settings like begging on the street or taking a metro. I got very intrigued by that; you never see a superhero in such situations. So that’s the first thing that I also tried out and it became my first post using AI. From there I started experimenting. That’s when I realized what this tool is capable of doing. I then started getting creative with my ideas, juxtapositioning thoughts that came to my mind, things that people wouldn’t usually imagine. That’s how it started.

Until last month it was very easy to identify. As you said, it couldn’t get a lot of human features right. For instance, it doesn’t know how to make fingers right; whether it’s five fingers or six. But it is a bit difficult to identify with landscapes. However, sometimes AI messes with the light’s position or the shadow’s direction.

It was easier to identify people with Midjourney V4 but last month the new version V5 was launched and it has improved a lot. I made the Slumdog Millionaires series using V5. If you check the hands and fingers, it’s pretty close. It has improved quite a lot with the new upgrade. Right now it has reached a point where it is difficult to identify. If I start an Instagram page with images of people generated from AI and tell everyone that they are real, they will believe; it’s got that good. And it’s only going to get better unless they put regulations on the level of perfection you can achieve with AI.

My major concern is that people don’t disclose that their images are created using AI and pass them off as real images. Only the people who are from this field or are technologically aware can tell the difference. The majority of us still do not know what AI is or that it exists. So they have no way of knowing whether a picture is real or not. A lot of people are doing this on purpose to spread fake news and misinformation. We still don’t have a solution for this because currently there is no regulation on who can use AI or how it should be used or what the implications will be if you use it to mislead people. It’s new for everyone.

Regarding people losing jobs, I don’t really buy into that. Technology will only progress. Some things will keep on replacing certain other things. Maybe mediocre artists will go out of business because AI can really give above-average results. I have received a lot of inquiries from movie script writers for mood boarding and scene creation which is normally done by traditional artists. While doing that I realized that AI cannot replace humans completely because there are certain things that you can’t create by typing in prompts, they require human creativity and intervention. I don’t know what will happen in the future but as of now, there is no real concern. If you are really good at your work, I don’t think AI can replace you.

No. It’s way beyond what AI is capable of right now. Video creation requires a lot of planning and direction. Right now, we do not have any technology that’s even close to achieving this. But you can upload an already existing video and manipulate it using AI. There is a program called stable diffusion that helps you do that.

There will be dramatic changes for sure. People have started realizing its power. Even traditional artists and photographers have started using it. A lot of these people were against AI three months ago but are using it now because they have understood that it’s the only way forward. If everyone is using new technology, you don’t want to be left out for too long. If you have no idea what’s happening or are averse to change, you are losing time.

With AI, you no longer need to hire a studio to generate backgrounds and make additions to your photographs. Or get professional help to create posters, flyers, or wedding invitations. Those jobs will soon become a thing of the past. AI has plenty to offer. Everyone should give it a try to at least know what is happening. Otherwise, it will get difficult to stay afloat.