Thoughts of a post-production artist in the days of AI

I don’t know how it is for you, but whenever I talk to friends or meet new people and explain what I do, the conversation eventually leads to a familiar question: “Did AI ruin your business?” (They don’t usually say it that directly, but that’s what they mean.)

So, what’s my answer? “Not really.” Sure, AI has changed a lot. I’ve probably lost some projects I’m not even aware of. But at the same time, it has opened up opportunities that never existed before. It’s helped me complete tasks in a fraction of the time they used to take. In some cases, I’ve been able to deliver projects without needing to outsource parts of the workflow.

Is it transforming the industry? Absolutely. Every week, I test new features, tools, or SaaS platforms to see if they can enhance my work. I subscribe to so many services that I keep a Notion list just to track them. But these tools allow me to create incredible visuals—things that would have been impossible a year or two ago. The main challenge is staying updated myself, and making sure my clients stay informed too.

Sometimes I tell clients something can’t be done within their budget (which they tend to reveal at the last minute, as if withholding it might get them a better deal). Then, a few weeks later, I stumble upon a new tool that makes it possible.

Client Expectations in the AI Era

I get a lot of questions about what’s possible with AI and what’s not. One common misconception: some people believe AI can do anything. They think they can hand over a poor-quality photo, and out will come a masterpiece that will leave the competition in the dust. Here are a few real-world cases I’ve dealt with recently:

“AI Can Replace 3D”

A client wanted a vehicle modeled in 3D and rendered from four angles. At the end of the project, he requested a walkthrough session—he wanted proof that I hadn’t used AI. My response: if you want a walkthrough, I’ll have to bill for that time. And frankly, AI isn’t capable yet of producing consistent multi-angle renders. But does it really matter whether I used 3D modeling or AI, as long as the end result is what you wanted? This is like insisting a photo must be shot on film rather than digital—what’s the point?

Amazon Product Visuals via AI

A client provided various product shots but couldn’t afford a proper location shoot—or even a 3D model. We used an AI approach, trained the system on his product, and created visuals in realistic environments. However, he expected perfect consistency. Even small inconsistencies became serious issues. Some things can be fixed in Photoshop, but others can’t. So if you like what you see, you either accept some imperfections or reject the image. AI isn’t a magic wand. It struggles especially with things like text and logos. Not everything can be fixed in post.

“If It's AI, It Should Be Cheap”

A client reached out asking if her fashion designs could be turned into photorealistic visuals. I assessed the situation and gave her a price range. (Most clients don’t want to share their budget upfront, hoping to get a “deal” after I reveal mine.) She accepted the range. I created a test image—something I rarely do—and adjusted it per her feedback. She said it looked fantastic. Then came the twist: she could only pay one-tenth of the quoted range. I stopped communication immediately. I assume she just wanted to see what was possible and gave the initial approval to get a sample.

Some people assume that because AI is involved, it must be cheap. That’s not how it works. At the time of writing, I pay for nearly ten different tools monthly—and that number might grow. I spend significant time learning and testing each one. I move visuals between platforms to get the exact results I want. And often, I still end up in Photoshop doing manual refinements.

If you think visuals should cost $5–10 per image, maybe it’s better if you subscribe to the tools and do it yourself. But if no one else has achieved what you're after—and I have—there’s a reason for that.

Final Thoughts

I’ve been working with images for over 25 years. Every scenario demands the right tool—there’s no one-size-fits-all. Every tool has strengths and weaknesses. When you need perfection, nothing beats a well-executed photo production. Sometimes you need to combine it with 3D. 3D works great for products, but not so much for organic subjects—even with recent AI advances. Our eyes are sharp; we can sense when something looks “too perfect” to be real.

AI has made huge leaps in just the last two or three years, but it still has a long way to go. I can often spot an AI-generated image from a distance. Maybe not everyone can, but a trained eye won’t miss it. That said, I use AI extensively, especially for moodboards and conceptual visual development.

So—what are your thoughts, whether you’re a client or a fellow image creator?