The Rise of AI, some more musings.
During a recent camera club talk I gave on Zoom for Camversation I was asked my thoughts on AI as I had mentioned it throughout. And I gave an honest answer at the time, namely that my thoughts were not fully formed. I have had several discussions since, mostly on Twitter and my thoughts are a lot more formed. Often for me, the process of writing a blog allows this to form even more fully so I thought I’d record my thoughts here. Feel free to add your comments below.
When I opened Photoshop after a recent trip to the Lake District I was greeted with a pop-up asking if I wanted to try the new ‘generative’; ‘tool’ I think I’ll call it. This allows you to make a selection or extend a border using the crop tool and add objects into your photograph. It’s a bit like a composite picture, which in itself is an art form in its own right.
And this is where the controversy begins. Should you? Ever? What about adding a sky that you shot 5 minutes earlier in the same session and blending exposures? For instance I was recently photographing Slater’s Bridge in the Lake District at around midday and there were people everywhere in and around the scene, enjoying the lovely weather. I had to wait some time for a clear shot and during that time I had several choices. Wait, which I did, shoot a nice sky and then wait until the people had moved and shoot the foreground and blend the two, shoot it and add in another cloud or add in a whole new sky, or remove the people using some kind of healing or clone tool. All of this can be done in seconds, but which should we choose? Below is the final image, what do you think I have chosen (answer at the bottom of the blog)
What is generative art in Photoshop?
First though, let me show you what I am talking about. If you haven’t seen it I have a short vlog, taken from the end of a Sunday video which shows what can be done to a landscape photograph. This is above and beyond just ordinary grading, dodging and burning. You can find the video here
I have to admit, this is all very clever. Some of the results are amazing. And some not so. Below are three images of the same scene from a sunset shoot at Black Crag looking into Little Langdale. It wasn’t the best sunset and for that reason I don’t think I was concentrating and the mid-ground needed to be less prominent in the composition. So I thought I would try the new Photoshop feature to sort this out.
And here are the results. Or at least some. The middle one is the original, on the left I wrote the prompt ‘still water reflecting a mackerel sky at sunset’. The one on the right I wrote ‘large bolder, granite’. It may be difficult to see on the device that you are viewing this on so let me give you my thoughts.
The two additions are fantastic. They would be easily good enough for instagram, Facebook etc and you would not be able to tell unless I’d told you. You definitely wouldn’t. They both fit the scene brilliantly, look completely natural and I would have been delighted had I found these features in the landscape. But on close examination and if I printed them here is where the problem lies, for now. A print would show the softness around the blend and a big print would show up the failings very easily.
That’s great, embrace it.
So, as you can see, there are a lot of possibilities but this has all left me feeling uneasy. You see, I feel that this is deceptive and will not be using it for any of my images. Why? Lot’s of reasons I suppose.
It’s not my work: you, my audience/followers/customers expect genuine images/stories/products that I have crafted and continue to craft from my years in photography. Adding something in from Adobe’s (makers of Photoshop) vast image library is using someone else’s hard work and it therefore isn’t mine. It’s a very small step, in my opinion, from adding in a water feature to making a whole new image from my office chair without setting step outside. Landscape photography is about the journey, the experience, the richness of light and sound of the natural world and the imagery, I hope, reflects that. If I become a digital artist I will let you know but I feel I need to state here and now that all of the images you see here on my website or in my YouTube videos will be genuine images shot with my decisions, technique and creative eye, using my own equipment.
Sincerity: I was lucky enough to have a photograph of mine exhibited some years ago at the National Theatre on the South Bank and, while I was looking at it proudly one Sunday morning, having persuaded/threatened/pleaded with my family to come down to London with me to see it, a man in a suit walked up beside me to look at it as well. I was very tempted to tell him that I had shot that image, it was all my hard work. Probably because I was expecting praise and to have my ego massaged. However, I restrained myself and instead simply asked what he thought of it. I was quite aghast when he replied ‘Yeah, it’s alright. But it’s all just photoshopped now mate ain’t it? Speechless I simply nodded and didn’t bore him with the 2am alarm, 2 hour journey I had undertaken to shoot it. But it highlighted a point even back then. The general public just don’t trust anything they see anymore.
Copyright: to a lesser extent, adding in other objects technically is not my copyright but rather Adobe’s and that won’t work for me either.
Genuine: finally I pride myself with being honest, it’s an important trait in any relationship be that personal or in business. If we lose honesty we lose trust and we then lose a whole raft of things that make modern day life fun and worthwhile. I could not, even though the images above have been improved by the additions, claim them to be my own when they are part mine and part AI. It just doesn’t work.
Dinosaur
So does that make me a luddite? I don’t think so. I tend to be an earlier adopter of technology and techniques. I produced panoramic by stitching images together early in 2006 before it was a thing (my phone can now do it in seconds), I was very quick to buy a drone and have enjoyed using them for imagery (I’m on my third). Indeed I was one of the first in my friendship circle to buy an iPhone, SatNav and all manner of other tech. The video above was produced as soon as I experienced the capabilities of AI and you will hear me say ‘Wow’ and be amazed by what appears but… as my mum has always said, ‘Just because you can, doesn’t mean you should.’ And for me as a landscape photographer who believes in capturing the landscape in all its natural wonder, this is a step too far.
I had the same discussion with myself many years ago with the healing/clone tool which is used to remove things from the landscape and I was initially dead against it, preferring to remove cans, rubbish etc on location but I am now happy to do that. So what’s different? Well it all comes down to the fact that what I am left with is still my own work and represents the landscape genuinely.
So what is the answer?
Already this has had an impact on photo libraries. Why would companies buy imagery from photo libraries when this actually gives them imagery for free? I’m not sure they would. Certainly the value of images licensed this way have reaped less and less value to the photographer who put the hard work in in the first place. There is such an influx of images from everywhere, mostly phones that the market is completely saturated. I said on a recent post on Twitter that photograph used to be a profession and it’s now a trade.
It’s also had an impact on competitions, organisers of which are now even more wary and worried about being caught out.
To solve this problem (more on whether it actually is in a minute) I think it should be pretty easy to have platforms where images are displayed read the exif data and identify if an image is 100% genuine. Selling prints though is a different matter.
We could also, as a group of professionals ask for some kind of regulation, professional body, accreditation and indeed professional bodies do exist. They could randomly ask for original files (RAW files) to be sent and inspected to ensure continued membership. We may end up with a trade logo or symbol that carries with it a guarantee of authenticity?
Does anyone actually care?
And this is a very important point. I have pointed out on various social media platforms when I think an image has been created with AI but this has been met with apathy in the main. The general public don’t really care to a large extent. People enjoy looking at excellent imagery and don’t in the main how it has been created. We live in a world where we scroll past thousand of images each week and the temptation to make something stand out by whatever means to grow a following is real.
But, from what I hear a lot clients of do appreciate authentic images And so do we as a landscape photography community I hope. I think we need to regain the trust of the public as professional landscape photographers and separate ourselves from the ‘composite art’ community (who I have the utmost respect for by the way) but that is going to be very difficult amongst the torrent of imagery out there.
We must care and we must continue to care and say so.
Final thoughts
Finally, I am just, at the time of writing, finishing my review of Luminar Neo which is a brilliant piece of software and has, as part of its suite of tools a way to add in skies. ‘Hypocrite!’ I hear you cry. Well… not really. It offers some fantastic tools other than that which will greatly benefit landscape photographers and I will hopefully show that in my video review. I won’t be adding skies in though. Like everything else in Photoshop there are features, tools etc that I have never used and some I use a lot. Generative AI will just be something that I will probably not end up using.
It’s all getting really clever but…remember, ‘Just because you can doesn’t mean you should.’ I’m going to keep my integrity and look on with interest at the rise of AI.
The answer to what I did with the Slater’s `Bridge image is that I used AI to add in the white clouds top left. Could you tell? Needless to say that image won’t be appearing here for sale.