Every shot I add to a post or article gets edited. Most of the time, they are not overly strong edits, shifts in contrast, lifting shadows, and of course cropping - as apparently it is impossible for me to shoot a shot straight. I edit in Adobe Lightroom, mostly on my phone, so it isn't really conducive for high end post-processing, but good enough for what I am doing, and the shots I am taking.
However, what I haven't tried until today, is AI. No, I am not putting AI-generated images in, but Lightroom does have a generative-AI "remove" function in its tools. And since I hadn't tried it, I thought I would see what kinds of results it would get with what would other wise be a pretty difficult (for me) edit in Photoshop, if doing it manually. But it seems, no one does anything themselves anymore, and just uses tool to gap-fill all of their skill-gaps.
For a remove, the AI has to predict what would be behind the object, filling it with something that it assumes will suit. While far from perfect, Lightroom gives some options in the refine section to choose which version suits. I didn't play with that much yet though, just used the default.
The first image was one I took at the beach yesterday of my wife, where there was a woman sitting behind in the background. It wasn't too bad, as it was on a strong blur, but it does create unnecessary clutter and breaks the horizon line. This is something I could fix in Photoshop, if I still subscribed for it. But instead, selecting the remove and blocking out what I wanted to take away with a little extra around the edges, it took about 20 seconds for it to get this result.
The second one was taken the day before yesterday at the cottage, again of my wife and this time brushing her hair. Fr me at least, this would have been a challenging edit, because it would have required rebuilding the scene piece by piece. And at least for this shot, there would be no way I would be putting that kind of time and effort into it. But again, for the sake of the experiment in Lightroom, it was a good choice to to try.
A few thirty seconds later.
There are five changes made and while the results are not perfect, it is pretty impressive that it puts in the corners and tries to make the shadows blend into the picture. Unless looking for the changes, it would definitely pass a once over glance, even though the angles are slightly wrong and the shadow unnatural. And of course, if I had refined it and played more, I am sure it would have been much better.
Because I know what has gone into the changes, I am not a big fan of them. Yeah, it would be fine for a cover image or something like that, but it is no longer completely mine and they no longer feel the same because of it. The way I capture photos is to reproduce what I am seeing, not for the image itself. Even when I edit, it is me doing the editing, me making the changes, me making the decisions on what is there, what is not, and what is good enough.
My heart is gone from these.
What it does point to though is how easy it is to make an image that is "real" to begin with, unreal in a realistic way. All those Instagram models using a hundred filters to smooth their skin, narrow their waist, brighten their eyes and whiten their teeth digitally, but are not that in reality. And I wonder, when they see themselves in images and then in the mirror, are they happy with the difference, or have they too, lost heart?
Generative-AI is only going to get better and people will use it more an more to fill not only their skill gaps, but also the gaps in their expectations, the gaps in their opinions of themselves, and the gaps in who they believe themselves to be. And the problem I predict is, the more we use AI to fill our gaps in self, the more gaps will form until there is very little left of who we are, and what we could have been.
Only Gaps.
Taraz
[ Gen1: Hive ]
Be part of the Hive discussion.
And you may be rewarded.