Artificial intelligence in photography


Artificial intelligence in photography is here, and commercial applications are already available that use it to generate text and image material for us. So what does this milestone mean for, say, a photographer or a designer?

In any case, things will change, or rather things have started to change. Willingly or unwillingly, we are entering a new era that will not only be a fad, but a milestone in visual communication. UI is becoming a part of our lives and seeping into every pore, in things like photograph, and this is especially visible. Of course, UI is about two concepts - fully generated image creation (like Midjurney) and UI advanced tools that will seep into photo editing programs in the future. Should we be concerned about the impact of UI on photography?

First of all, let's ask ourselves: Why and what is photography? Is it a documentary, reportage depiction of the moment? Is it a product photo or a marketing concept visualization? Or is it about telling the artist's story through the prism of photography? I think everything at the same time. So how does UI affect these foundations? Well, I will try to answer these 3 questions from my point of view:


Capturing a moment or a documentary photo.

We can't talk about the excessive influence of UI on this genre here. UI cannot generate an image because it does not have a database of identical situations and therefore cannot be documentary by definition. We photograph what is happening while it is still being created and everything is still a blank slate, and any subsequent manipulation destroys credibility. Reportage photographers don't have much to fear from UI, but amateurs with mobile phones, of which there are always plenty everywhere, have eaten away their bread anyway.


Product photography and commercial use.

A base is required to generate an image of a product. In the database from which the UI draws image material, there is probably no photo of the XY product that the client would present. We can create a background and use the UI as an effect tool, but we will not be able to make the entire photo by ourselves. In commercial photography, the UI will be reduced to the creation of "atmosphere" or supporting graphics. It will probably also make it easier for us to process to the desired effect, which is sometimes painstaking work in Photoshop. I'm talking about the case when the UI is built into the tools for processing an already created photo. However, when it comes to fully or partially generated images with the help of UI, the principle of content will probably be similar to that of stock photography - as long as it is about providing a wider picture or an instant story, supporting news with symbolic photos, providing generic visual support for the rest of the content... this works. There are very good and likeable photos, and there are a lot of them. But if we want a photo of a specific content, we have to do it ourselves and help ourselves with less advanced algorithms, such as tools in Photoshop, etc... The UI will, in my opinion, be used more than once as an advanced tool and will in no way replace the photographer in the creation of fully defined content . Also, more complex visualizations of the final product will become more the domain of the designer or creative, who will upgrade the visual material using UI.


Telling the artist's story through photography

In my opinion, UI will have the greatest impact on this part. It will be difficult to blur the line between photography and concept. After all, this is about the use of photography in the context of general visual art, and its role is therefore reduced to the acquisition of raw material, which the photographer (artist) then transforms into the desired project or visual product. In my opinion, the development of the UI will also go more in this direction, so that it will be able to understand the artist's requirements and place them correctly in the creation. So according to the plan and idea, which she is not yet able to do, despite the impressiveness and admiration of the generated images. These are random approximations of the desired, which is otherwise fun and interesting, but the question is if this is exactly what we are looking for. But in general, this whole "art" category was not photography before, it just used photography as raw material and made something third out of it or with it. Without our imagination and clear expectations, there will be no good results here either.


Legal formal issues

More pressing to me are the legal issues regarding the authorship of images that are completely generated by the UI and the authorship of images that the UI uses as a database. The authorship of the image generated by the UI tool cannot be in the domain of the algorithm (the author can only be a person), nor in the domain of the owner of the software that uses the UI for this purpose. The reason is simply that without my initial wishes, guidelines, instructions, expectations, the picture cannot be created by itself. And from the moment the image was generated at my request, it is MY author's work, at least in terms of idea and general purpose. Anyway, UI never makes the same products despite the identical input data, so the combination of my idea and the tool is unique. But it is not art if it is also partly equated with the craftsmanship of the artist. There is no such component here.

Another issue is the base that the UI uses. This can be compared to a photographer or an author who, after looking at thousands of photos, pictures, etc., gets inspiration, draws something from there, something from here and then makes something third. That's just how it works. If the UI uses a similar principle, there is nothing controversial as long as it does not use images or photos as micro-pieces of the puzzle. This is a bit controversial, because it is no longer about ideas, but the use of the original work, maybe just a piece. Here, copyright law will have to go a step further and enable the protection of works whose authors do not allow their use for this purpose. UI image generators should have indelible and traceable information in the image itself that it is an artificially generated product.

I think that, after all, we don't need to be so afraid of the UI, to think of it as a bav-bav that will take away all our work, a ne'er-do-well who will do everything cheaper than us, etc...


Baw-baw or not?

I think we need to take UI more as a new tool to help with processing, finding new ideas, facilitating some production processes. The introduction of digital photography also brought changes that went in this direction. And of course, it is cheaper and easier to retouch a photo on the screen than with a brush (although it is true that we need new skills for this). There will always be puritans who swear by photography without influence or processing, but the processor in our cameras already contains algorithms that beautify our image. The development of the UI will probably go more in this direction, so that the tools become more powerful and accurate. If I want, say, no clouds in my photo, the UI will remove them by itself and will know very well what I want from it. It's the idea that counts.


Tomaž Berčič