Should photographers worry about AI in photography?

Artificial intelligence in photography shows tremendous growth. We’re even seeing our social media feeds being flooded with AI-created images. Just last month, we saw an AI-generated image win first place in an art competition. We’ve also seen Getty Images move away from AI-generated imagery. I agree with that decision, but we also need to look at the broader landscape because AI art quite simply isn’t going away.

The copyright issue

Looking at Getty Images’ position first, why did they do that? They’re not the only platform implementing this position – Newsgrounds, FurAffinity, UnSpalsh, and PurplePort have done the same. A very important reason for this decision probably comes from copyright. It’s not that simple… A machine owned by Company A trains on Set B and outputs an image according to Person C’s input. Who owns the copyright? This is new territory and I can see lawyers banging their heads against the wall with it. You might want to grab some popcorn for the upcoming lawsuits.

If I were an attorney at Getty Images, I just wouldn’t want that headache. At the end of the day, Getty Images is all about making money for themselves first (and for the contributors second). Alongside this, they want to offer a simple and effective customer experience. Getty is probably big enough to do all of that and hold its market without having AI art on its platform. But what about other aspects of AI in photography?

The artistic debate

Dall-E 2, MidJourney, Craiyon, Stable Diffusion and the other vendors that generate AI art based on text input are fun. But there’s an argument that they’re seriously blurring the lines of reality. We can create photorealistic scenes that could be confused with real places or situations. There is also an argument that creating AI art does not require sufficient human input, nor does it require sufficient investment in skills or time. I typed’AI in photography‘ in Craiyon and got this result:

Using this extremely simple, short sentence to give me nine images didn’t require much effort, so this argument is true. Just to add, I’m a little worried about what’s going on in the bottom right image. Continuing, I said at the beginning of this post that AI art is not going away, so let’s see 1 – how we can use it in our photography and 2 – see what concerns we need to overcome.

advertisement

Artificial intelligence is the result of machine learning. In short, this means that a computer is fed data in the form of images – millions of them. The code in the computer is based on a training model, and the machine is trained to learn what is right and what is wrong in those images. The output coming from the AI ​​in this case is an image or a modification of an image. It’s the modification of the image that I want to focus on for now.

What about the process?

We see AI being used by Adobe in Photoshop. Mainly in the form of neural filters. Adobe’s AI has a name: Adobe Sensei. Sensei is not only used in Photoshop, but in many apps in the Creative Cloud and Experience Cloud. Adobe Sensei was announced at Adobe MAX in 2016, and as such, machine learning capabilities have been around for six years to refine our workflows. Similar AI technology is implemented in other programs such as Luminar Neo, Topaz Photo AI and the ON1 AI series. I’ve seen countless comments about how AI will ruin photography on social media, but let’s get on board with this implementation of AI. Do you remember? We’ve seen the same hesitation in moving from film to digital and then moving from DSLR to mirrorless. Even almost using Photoshop has not yet been fully validated. There are understandable concerns about introducing AI to post-production. What I have seen from testing and reviewing all these programs and being an Adobe Influencer is that they all make our lives easier and are constantly evolving their skills.

For example, the processes we learned for our image editing in Adobe Photoshop are outdated in some cases. The AI ​​software carries out these processes for us. This means that many of the things we have learned will become obsolete. And fast. But here’s the thing: the AI ​​makes it faster and, in many cases, better. We no longer need to exert massive effort to complete simple tasks. We can just press a button and let the AI ​​do everything for us. This creates a lot of freedom for creativity and helps us with insanely good sharpening and pixel upscaling. Check out this example of what Topaz Photo AI can do to sharpen this puffin I shot in Iceland. You can see how cropped it is by checking the original image at the top right.

advertisement

Shift up a gear with Luminar Neo. Watch as the AI ​​analyzes what’s in my photo and makes suggestions on what improvements to make above and gives me access to all the other features of the program. We have long known about the ability of software to recognize elements, faces and parts of images. This is a perfect application of this technology. Advancements like this one from Luminar Neo are great tools to complement our workflow.

So here I am. The question asked was: ‘Should we be worried about AI in photography?‘ My answer is no. We are in a position where AI is an industry in its own right and its place in our industry is becoming more established.

We need to embrace the positives and (oh, the irony) educate the industry on them, which is a valuable addition to our workflow. When it comes to workflows (e.g. Adobe Sensei, Luminar AI, Topaz AI, ON1 AI and others) I embrace them with open arms. They save me time and reserve my mental fuel for the creative situations where it’s better placed. As for fully AI generated images, I feel like there is a time and place for their use. If you had asked me if AI-generated art alone had a place in photography, I would have said no, but then I saw what Ramtin Kazemi did. Follow this insta trust me.

It may have a place after all – only time will tell for sure. Using AI to help the process rather than creating an image from scratch is certainly a big help.

What are you saying? Should photographers embrace AI?

ps On a completely different note, you might want to read this book.