The Reading of a Book

“The Quiet Reading-A-Book-Ness Of Reading A Book”

Click image to view on Flickr

• • •

Or: “How I found legitimate use for AI in my creative process”.

This is a photo that I took of a fellow passenger on my commute a handful years ago (2019), and it’s one that I’m quite pleased with. I tried to capture some of the peaceful mood and atmosphere surrounding someone who is experiencing the intense “reading-a-book-ness” of quietly reading a book, firmly undisturbed and undistracted by the world around—including rogue photographers—and I feel that I may have managed just that.

• • •

But there’s a backstory, and for that I had to get a little help.

Some of the keen-eyed among you may have noticed that something is slightly out of tune. For the simple reason that I never got the chance (read: “had the nerve”) to ask her permission to publish the photo, and, by ethical standards, I cannot publish her likeness without her consent, and I really, really, wanted to publish the image, I had to find a workaround… and four years after taking the photo, that workaround arrived in the shape of AI.

Now, I should interject that I’ve been quite skeptical and critical towards AI-generated content. Though this applies mainly to AI-generated text and images being presented as ^{[insert air-quotes]} “the real thing”, of which there is far too much going around, and people are eating it up like ice cream on a hot day. With this image, however, I had a quite different approach, using AI simply as a tool along with my own creative efforts, so I think that I am pretty much on dry land with that.

I rejected the idea of blurring out her face, as that would completely detract from the mood I wanted to convey. So instead of presenting the original photo (which is not an option) I’ve anonymised her by using an AI image generator (which weren’t commonly available back then) to render her a new face, based on her original one, which approximately (but not too closely) resembles her, without being identifiable as her. I then carefully blended the face (and only the face) into the image, taking care to match the lighting, colour, and texture. I’m sure that it could have been done better, perhaps more photorealistic, but I’m still learning new AI skills, and I may give it another try at a later time. So all in all, only the face was altered, the rest of the image is exactly as it was, and I am very happy with the result.

📷 iPhone SE (2016), using the Lightroom Mobile app (for RAW image), edited in Lightroom Mobile and Photoshop Mobile on iPad.

• • •