I’ll be honest – I’ve been putting off writing this one for a while. Not because I don’t have opinions on it (I very much do), but because it’s one of those topics where everyone seems to have already staked out their position before the conversation even starts. AI = bad, photography = pure, end of discussion. Or alternatively: AI = just another tool, get over it. The trickle of discussion about AI over the last few years has become a full-on torrent.
The reality, as ever, is a bit more complicated. So here’s my attempt at an honest take – as someone who loves the craft, spends a frankly irresponsible amount of time checking weather apps, and has a lot of thoughts about what makes a landscape photograph worth making in the first place.
The genuinely useful stuff
Let’s start with the positives, because there are some real ones.
AI-powered noise reduction has been an absolute game changer. Anyone who shoots the Milky Way, or drags themselves out to the Somerset Levels on a -3 degrees January morning for a misty sunrise (guilty, repeatedly), will know that pushing your ISO into uncomfortable territory has always come with a cost. That grainy, crunchy quality that would appear the moment you zoomed in and immediately ruin your day. I have so many noisy, lower quality drone images that have now become possible to turn into large prints. And yes, a tripod helps – but there’s only so much it can do when the light is that low.
Tools like Topaz Photo AI and Adobe’s AI denoise in Lightroom have genuinely changed what’s possible (the recent denoise upgrade in Lightroom is ridiculous). Images that I would previously have binned – or quietly filed away in shame – can now be rescued into something printable. For night photography especially, it’s hard to overstate how much difference this makes. The noise reduction in these tools is operating at a level that simply wasn’t available a few years ago.
Lightroom’s generative upscale feature is another one worth mentioning. It uses AI to significantly increase the resolution of an image – meaning files that might previously have only been suitable for smaller prints can now be turned into something much larger without losing quality. Useful if you’re sitting on older images from earlier cameras, or if you want to offer bigger print sizes from your existing catalogue.
Subject masking has come a long way too. Selecting a tree line against a complex sky, or isolating a misty foreground for targeted processing – stuff that used to take a fair bit of careful, tedious work in Photoshop – now takes seconds and is often more accurate than what you’d achieve manually.
These tools are serving the image you already made. The alarm still went off at 4am. You still drove to the location, set up in the dark, and waited. All of that is completely unchanged. The AI is just helping you get more out of the raw file at the end of it.
It’s not just post-processing either. AI has become genuinely useful for planning shoots too. Tools like ChatGPT are surprisingly good at suggesting new locations to explore based on what you’re looking for – whether that’s a misty valley within an hour of Bristol, or a coastal spot that works at low tide in winter. You can also ask for camera setting starting points for a specific scenario, which is handy if you’re venturing into unfamiliar territory like astrophotography or long exposure work.
It’s also worth remembering that AI has been quietly embedded in our photography for longer than we might think. The autofocus systems in modern cameras, the computational photography that makes a phone image look that good in low light – that’s all AI working away in the background. Isn’t it funny how iPhone photos always seem to be perfectly exposed… We’ve been using it without really calling it that.
The bit that’s more complicated
Here’s where I start to get a bit uneasy.
AI image generators – Midjourney, Adobe Firefly’s generative functions, ChatGPT’s image tools, and the rest – can produce landscape imagery that, at small sizes and to non-photographers, is genuinely difficult to distinguish from the real thing. A dramatic Glastonbury Tor at sunrise, mist rolling across the levels, golden light. Generated in seconds. For free. Social media is awash with AI images and videos – Facebook in particular is a shocker for it.
For stock photography, particularly the kind of broadly atmospheric, non-specific imagery that used to form a reliable income stream for a lot of photographers, the impact is already being felt. If a client just needs “moody British countryside, autumnal vibes” for a website header and doesn’t much care whether it was taken by an actual person standing in an actual field… That’s a problem.
What it doesn’t replace is work with a specific identity. Images tied to a real place that someone actually wants to put on their wall – a print of the Clifton Suspension Bridge in morning mist, or Glastonbury Tor with the kind of light that only happens a handful of times a year – aren’t really competing with generated images, because people buying those prints know exactly what they want and why. The “this was real, I was there, this light actually happened” element is the whole point. Can an AI-generated image really tell a story? Will there be an emotional attachment to it? The photos I am most proud of are the ones that I feel a connection to. You can’t connect with an image that has no soul.
But generic work? Work without a strong sense of place or voice? That’s where the pressure is coming from, and it’s not going away.
The dodgy grey area
Then there’s the thing that really bothers me.
Some photographers are using generative AI not to create images from scratch, but to add elements into otherwise-photographed scenes. Better clouds. Animals that weren’t there. A clearing mist that the actual morning stubbornly refused to provide. Completely switching out a background. Or one I’ve seen a few times recently – a winter wonderland where there wasn’t one at all. Sometimes this is disclosed. Often it isn’t.
I think this matters, and not just for competitive reasons (though competition entries are an obvious flashpoint – most now ask you to declare AI use – not that that stops people). The reason I started photographing mist was because of genuinely misty mornings. The whole point is that it’s real. There’s a reason that image of the Tor in the mist means something – because the Tor was actually in the mist, and I actually got up before dawn to be there. If you add the mist in Photoshop, or generate it with AI, you’ve made something that looks similar but is fundamentally a different thing.
I’m not naive enough to think the line between “photography” and “digital art” has ever been perfectly clean – we’ve all dodged, burned, used selective masks, and nudged colours into territory they weren’t quite at in real life. But there’s a difference between processing the light that was there and generating light that wasn’t.
The industry hasn’t fully caught up with where to draw that line. While it does, I think it’s worth being very explicit about your own process – both for your own integrity and because transparency is increasingly what builds trust with an audience.
Will it make us worse photographers?
Some photographers worry that AI post-processing tools, by handling what were once skilled operations, will slow down the development of craft – particularly for people just starting out.
There’s something to this. Understanding why noise behaves differently at different ISOs, or why a luminosity mask produces a more natural-looking result than a simple selection – that kind of understanding feeds into every decision you make at every stage. If the tool just does it for you before you’ve ever thought about why, you might miss something.
That said – I’ve seen this argument made at every stage of photography’s history. Film photographers said it about digital. Darkroom printers said it about Lightroom. Each time, the photographers who genuinely cared about the craft continued to understand the principles underneath the tools. The ones who just wanted to press buttons continued to press buttons. AI doesn’t really change that equation.
What it can’t do
Here’s the thing, though: AI cannot wait for the light.
It cannot make the decision to get up at 4:30am on a Tuesday in November because the forecast shows a small window of mist and the moon is in the right position. It cannot stand on a hill above Bristol in the dark, cold, and quiet, wondering if this is going to be the morning that makes it all worth it (and occasionally it very much is, and occasionally you drive home having seen absolutely nothing of note and spend the rest of the day scowling).
The images I’m most proud of – the ones in my print shop that I actually want to sell, the ones that have meant something to people – came from that process. A specific place, a specific morning, specific conditions that aligned briefly and may never align in quite that way again. AI can produce something that looks similar. It cannot produce that.
For photographers whose work is rooted in a real place and a genuine relationship with it, I think the foundations are pretty solid. The photographers who know the Somerset Levels well enough to know where to stand when the mist rolls in a certain way, or who have spent enough time with Bristol’s light to know which mornings are worth setting the alarm for – that knowledge and that presence is the thing that matters. AI doesn’t have it.
Where that leaves us
I don’t think landscape photography is dying. I don’t even think AI is the biggest threat facing it – the commoditisation of technically-competent photography has been happening for a long time, and the photographers who have built a clear identity and audience have been fine through all of it.
What I do think is that it’s worth being clear-eyed about where the pressure is coming from, using the tools that genuinely help (and some of them really do), being transparent about process, and – most importantly – continuing to make work that couldn’t have been generated. Work that required being there. In a way, that’s no bad thing – it pushes us towards originality.
The algorithm doesn’t do freezing cold mornings. That’s still ours.
If you’ve enjoyed this post and want to see the kind of images that require horrible alarm clock times, you can browse the print shop here – or follow along on Instagram where the mist updates are essentially a lifestyle at this point.
