The advancement of AI in photography and videography spheres has come more rapidly than many people are comfortable with. But there's one issue that hasn't been discussed that could have major repercussions for the industry.
Along my photography journey, I always recognized that editing was equally, if not more important than actually taking photographs. To that end, I spent a lot more time honing my skills in front of a computer than I did behind a camera. Don't get me wrong, I absolutely love photography, and nothing gives me more joy than getting out in nature with just me, my gear, and the landscape in front of me. That said, I always felt that post-production was really where I could separate myself from the masses, so that's how I tried to keep ahead of the pack. Turns out it was a forlorn task, as the evolution of AI has quickly overtaken me. A lot has been made of how AI is disrupting the industry, but one topic has not been examined yet, and it's one that could have major ramifications.
That brings us to this great video by Taylor Jackson, in which he discusses the issue at hand: that AI has become so sophisticated and realistic that beginner photographers thinking about buying their first cameras could be so intimidated by the quality of AI and its possibilities that they don't bother purchasing gear or starting their journeys. It makes sense when you think about what you can already do with AI programs like Midjourney, for example. If Jackson is correct in his thinking, then this could have a massive impact on sales for the big makers like Canon and Sony. What do you think? Have you tried AI yet? Let me know in the comments below.
Apple already uses a lot of AI for the iPhone camera. The stuff it does seems to be the basic set of things that a lot of photographers do, either in the field on in post.
The things I've seen it do include:
- adjusting a polarizing filter in front of the lens
- bumping up the saturation of common elements like the sky
- using texture synthesis (i.e. "content-aware fill" in shadowed areas (with varying degrees of success)
The thing about AI is that it's only as good as what you've already trained it to do. More broadly, it can do what others have done before. That's why it's so good at writing crappy essays for English class. It doesn't supplant real creativity or originality. Don't expect any deep revelations from it.
A pointless point I thought. AI is not going to stop people starting to take photos. It’s big issue is wiping out a lot of commercial photography used in advertising. I think we will get completely sick of it. There is no skill in it. Clunky interface now but it will get easier to use. After an initial explosion of use people will reject it. I think it will always be missing something. I think if something isnt limited it’s lesser in value. When everybody can produce AI images no one will value them. People need skilled creative outlets not simple ones.
I don’t think commercial photography is the only subset that should be worried. And adding to that, in your last sentence, when you refer to “people”, I assume you mean us, the photographers? That might be true for the producers like us, but for the consumers, I don’t think so.
If someone can produce an image to their pleasing with a few clicks of a button on their bedroom PC, why would they pay $100s or $1000s for the services of a photographer? That’s the dichotomy: how we, the producers view and interact with AI as it develops, and how the consumers who have to stump up the cash view it, if it’s a viable, almost free alternative to professional photographic work….
I think camera manufacturers are fine. At this point, camera sales seem to have levelled off, so even smartphones are no longer negatively impacting the market. Camera manufacturers have never put serious effort into creating tools for the younger generation of creators, so I think there's not a lot of overlap between would-be camera buyers and AI creators.
I hope you’re right, but having seriously played with AI for the last week or so, I have to say I’m absolutely stunned at what it can do. Once that knowledge and understanding filters down and through to younger generations, I wonder if they’ll think “why buy a camera? I have my phone and AI”…
I see people who want to take photos themselves of the world from their perspective - the world that they see and experience - as a different market than those who want to use a computer to create images that they need for some purpose. I don't even really see an overlap between these two very different types of people.
Commercial photography, involving the creation of images for advertising and promotion of products and services is such a wee wee tiny little part of photography that losing that super tiny niche genre won't have any noticeable impact on camera and lens sales. I mean, I bet less than 2 percent of all camera bodies and lenses are sold specifically to commercial photographers. So not enough of a part of the overall camera market to have any impact at all.
I think that if all you want is a good looking, well-exposed snapshot, then yes, computational photography will do it for you. This is another aspect of, “Can I use my phone camera to shoot my sister’s wedding?” The answer is yes you can but your results, computational or AI there or not, will be dismal compared to pro.
We’re well past the place in imaging where it took skill to get a photograph properly focused and exposed. What’s missing from being at that point to making a great photograph is composition and having an eye for something that will be worth the effort of careful composition.
Also, like those two items, skilled post-processing moves beyond what any computational algorithm can manage because good post is not rule based but founded in artistry.
Photography, like any art, is less technically demanding now than before. That’s true, but the base requirement of the photographer showing the viewer something they’ve never seen before, even in a mundane scene, that will never fall to AI.
Don’t take this the wrong way, because it’s not said in sarcasm. Genuine question: have you played around with AI programs?
I have been doing so for the last week or so and … holy smoke. The idea of composition, post-production skills etc…AI has all that covered and then some (I’m talking landscape photography). It’s creating images that people can’t create - because they don’t exist (unless you’re a composite photographer).
A lot of work still to do with people and portraits, but in fantasy, landscape, stock imagery etc, it’s cutting a swathe through the naysayers’ doubts, in my initial opinion
Iain Stanley wrote,
"It’s creating images that people can’t create - because they don’t exist "
And that right there is why AI will not have much impact on photography as a whole. Because the VAST majority of photographers take photos so they can preserve a memory of what they saw and experienced in real life. Their Aunt Edna on her 90th birthday. The rainbow they saw while driving home from church last week. Their kid in his soccer uniform before a big game. The delicious salmon dinner they ordered at a good restaurant. They take photos to record what was really a part of their everyday lives, not to create images of scenes and object that don't actually exist. I've no doubt this accounts for like 99.9% of all photos taken worldwide every day.
Only a very tiny subset of people use photography to create images that are for "art" or for some commercial purpose. And these are the only types of images in which AI will supplant a significant portion of actual photography.
Your last point is the crux of the issue: if photographers want to make money from photography, they need people/companies etc willing to buy their images. If AI can create those images quickly, easily, and cheaply, (pro) photographers become redundant.
You’re right about preserving memories and moments etc, but you don’t need to pay for any of those moments, outside weddings etc. Everyone has a perfectly capable camera in their pockets for taking a lovely pic with Aunt Edna.
I've been working with AI imaging for almost 6 months and seen incredible improvements as numerous models have gone through their iterations. I've worked with multiple versions of MidJourney, Stable Diffusion, DALL-E, Google's uninspired Imagen, and Crayon. Iain, you are absolutely correct that AI can produce beautiful images, with lovely composition, excellent lighting, exquisite colr balance, and gorgeous subjects.Better than the majority of my photos I must admit.
That said, aside from ending up with a pretty image, that I didn't create, it has become more of a personal puzzle to solve. In the end, it isn't very satisfying.
If your idea of photography is to simply produce a nice image, then sure, AI is probably enough in many cases. I don't mean this as a snotty comment. If you enjoy it, go ahead and enjoy it. You just need to solve the puzzle of finding which words to use to allow the AI to come up with something you like, but it will never be exactly what you envisioned. And of course, each AI model requires different phraseology to achieve similar results. I'm not saying this is right or that it is wrong. If you are having fun and like the results, I say go for it.
I am speaking only for myself here, but it has gotten a little boring for me over the past 6 months, not that I am not still obsessed with it. I still play nearly everyday. I am now experimenting with the combination of AI and my own photography. For me, the enjoyment of photography comes from not only the image, but the journey, the planning, the thoughtful geometry, the unexpected things that I see that I wasn't expecting when I set out to shoot a photo or group of photos. I can tell you that I would never in a million years pay for an AI generated image. It's too easy. I will cheerfully pay hard earned cash for an exquisite photo. I know what went into it. I know that the photographer controlled everything about it. I know it is a part of him/her.
I agree with pretty much everything. The difference is the lens we’re looking through. You’re seeing it from your own POV, naturally. I’m looking at it from the POV of a potential customer/purchaser.
We, as the creatives who love photography and produce images we love, will never tire of the craft from an individual, artistic outlet POV. But people shelling out money? They go with the cheapest option that gives them what they want. That’ll be AI every day of the week with the way it’s going.
I’ve only seriously delved into it in the last week or so, and like you, have been experimenting with different command combinations. Fascinating, thrilling, stunning, and scary is how I’d describe it thus far.
That’s a very valid point. The fear I have is when AI becomes so universally accepted within society and reaches critical mass where everyone knows how to use it, just like today’s smartphones. At that time, will everyone be so familiar with AI and its capabilities that it won’t be just the marketing agencies going cheap?
I have done a LOT of AI imaging and it is sometimes frustrating that I can easily write prompts that will eventually produce an image often better than my best through the lens images. Not only that, but there is no hiking to that special spot at 5am or waiting for perfect light, or...I can stay in my pajamas all day if I want.That said, speaking for myself, nothing can replace the enjoyment of either planning a shoot, or spontaneous street photography when a final product is something I am happy with.
I agree there may be danger for new folks who have not had the experience photography brings, to think about skipping the work, not to mention the expense of photography. Ultimately I expect that there may be a temporary dip in photo gear sales as budding artists experiment with AI. Although AI is moving fast, it is still impossible for the AI to read your mind and produce exactly what you envision.
I believe that the net result over time will not affect sales much. Instead, we will have a new medium for young artists to explore. This is a good thing. An artistic soul needs an artistic outlet. Who are we to decide what that outlet is. My belief is that for most, lacking total control over the final image will push the artists to medias where they have more finite control. AI can introduce the masses to art, but I don't believe artists will be satisfied ultimately.
What I have found interesting during my journey is a combination of AI imaging and my own photography. This can go both directions. Starting with a photo and making edits with AI, or starting with AI images and either editing in Photoshop or compositing with photos in Photoshop. Possibilities are endless.