The Blood Moon and the Banana: The Battle for the Soul of Photography

Three things happened the other week that have had a profound impact on my thinking about photography, and especially its future. This was on three levels: the first is the personal; the second is the technical; and the third is the professional.

The three things: First, the lunar eclipse last Sunday. This celestial phenomenon resonated with Kenyans in a way that many such phenomena haven’t for a long while. Blood moons happen far more often than you would imagine (every few months in most instances). But this one garnered an inordinate amount of attention for various reasons: it was the longest total lunar eclipse in a long while, and as such, many jumped on the opportunity to hastily try and connect a tourism angle. Also (and perhaps depressingly expected) was the Biblical end-times hooey that cropped up (as it always crops up whenever something unusual happens).

Whatever the reason, many of us had our eyes turned heavenward that Sunday night. We fought dust and cloudy skies to stay up and catch what was, truly speaking, a spectacular lunar show. In my case, the whole family was out on the balcony casting their eyes on the eastern horizon. The three hours between the first shadows on the full moon, through the faint crimson eclipse, to its re-emergence just before ten o’clock, were worth staying up far beyond bedtime on a Sunday night.

My photo of the moon, just as the earth's shadow began to cross over from the bottom left

And of course, that brought out everyone and their baby sister with a phone or a camera (which, nowadays, usually means the same thing). Many, many millions of pictures were made, with many of these shared in real time. These ranged from the mundane to the ethereal. But while some pictures were carefully constructed artefacts, with the right exposure levels, artfully composed and cropped for maximum artistic effect, many were the simple result of a mobile phone thrust in the general direction of the sky, a button pressed, and the result quickly shared on the socials. And here is the dismaying fact: the result of these two approaches is starting to converge.

As I said on my Instagram, taking good pictures of the moon is fiendishly difficult. It is a distant object, unusually bright against its surroundings, and moving surprisingly fast through the cosmos. I tried to take a set of high dynamic range images the following night, so that I could expose for the full moon, as well as the surrounding clouds, in order to have a good approximation of what the eye sees, and it was almost impossible. Even in a 1/3 of a second exposure, with the camera on a tripod, the moon moves perceptibly, so much so that it results in a blurred image. And this is not accounting for the movement of the clouds, which scud across the sky (and across the moon) in a way that is difficult to freeze in a photograph. This then leads the photographer to make certain compromises in terms of exposure levels, shutter speeds and the like.

What then accounts for so many near-perfect moon pictures from last Sunday? It has been documented before: certain mobile phone manufacturers go as far as replacing the (real) blurred image you take with a simulacrum. A fake. And even for those manufacturers that do not, there is a great deal of computational firepower deployed in order to turn our desultory shutter presses into acceptable, and often exceptional, pictures. The approach involves removing camera shake (hence blurriness), resolving pixelation due to zooming, white balancing, resolving dynamic range, and much more. The result is a picture (one cannot realistically call it a photograph) that is ‘better’ than real.

Now, I am not for once saying that all photos should be the result of professional-grade skills. After all, even serious photographers use computational tools to improve their work – anything from enhancing colour, to removing haze, to eliminating unwanted objects. But in all this, there is an implicit (and often explicit) understanding: tampering with an image beyond a certain level is unacceptable, or is acceptable within parameters such as doing it for artistic purposes. Photographers who try to pass off the fake as real are shunned. At the same time, though, most of us take snapshots purely for documentary purposes – a baby’s first step, a hike in the Aberdares, an artistically plated meal – photos which will never be entered into a competition. Therefore, phone tools that enhance lighting and skin tone, or help to compensate for under- or over-lit scenery, are OK. But it does mean that our memories are all-too-often airbrushed ones. Even if you were to stand with the photographer at that exact spot and in the exact moment the shutter is pressed, you will not (and cannot) see what they eventually share as the results of their photography.

And all this is before we get to the serious stuff, which is the second point. In the last couple of years, the very soul of photography has been put to the sword. Extraordinarily powerful software has been unleashed into the world; software that imperils the very notion of what photography means, and calls into question why anyone should bother to even pick up a camera. I am, obviously, speaking about image-generation software, or what has been cast into the broad rubric of artificial intelligence. Just a few short months ago, it was easy to dismiss these tools, especially when it came to generating images of people. Skin tones were too smooth. The ‘lighting’ was off. Soulless eyes. Six fingers on one hand. Disembodied body parts. But these tools were advancing at an incredible speed.

Some days ago, my friend Moses Kemibaro shared one of them: Google’s new Nano Banana AI Image Generation tool. This is no longer even about typing in prompts (‘please generate an image of an elephant in a forest surrounded by fairies’). It is a tool that combines real images (for example a headshot) with a scene. So our dear Moses has been a Formula One driver (wearing Lewis Hamilton’s race helmet, no less), a very handsome mubaba, and the Arsenal’s goalkeeper. My other friend Daudi Were has not been left out. He is now a member of the Springboks coaching staff.

Of course, these may be used as nothing more than amusements. Do you want to see what you would look like with a sixpack? Or at the Met Gala wearing a befeathered fashion piece? You will waste many fun minutes putting these tools to that use.

Music aficionados will remember the song ‘Unforgettable’, in which Natalie Cole sang a duet with her famous crooner father Nat ‘King’ Cole. Two indescribably beautiful voices  brought together in a studio production a quarter-century after the father’s death. This was as long ago as 1991. Tools like Nano Banana can (with the full understanding of the user and the audience) be used to create similar heartwarming scenes that never existed. A child whose father died far too young? You can now put the two at the same dinner table as young adults, sharing a tender moment.

However, it does not take much imagination to see how such powerful tools can be put to malign, even evil-intentioned, use. We have obviously seen nude photographs being used to harass or to destroy people’s lives and careers. And this is when the nudes in question were real (if ill-advised). How much more damage when fake images, but which are indistinguishable from real ones, begin being created and shared? (And this has begun, by the way – deepfakes, as they are known, are a regular feature of the cesspool in the darker parts of the web). There is more: a politician being depicted consorting with the ‘enemy’. A person placed at the scene of a crime or an atrocity when they were never there. When the stakes are sky-high, people will not hesitate to use readily-available tools to reach their malevolent aims.

We like to comfort ourselves by saying that we can easily detect AI-generated images (the floating six fingers). But that is precisely the point: the tools are now good enough to be indistinguishable from legitimate photographs. Also, we tend to forget that more than 99% of the world’s still and video images are not created or shared in high definition, where the audience goes pixel-peeping to detect any tampering. We share low-quality images across social media, quickly consuming, forwarding and discarding hundreds of images in a day. If a particular picture or video reinforces particular biases and prejudices, the real-world consequences may already begin to take place even as the experts scream into the void about the counterfeit nature of the images.

We are soon getting to the point where believing the evidence of our eyes becomes a sucker’s game.

On to the third thing from last week, which is also to do with Nano Banana and similar tools. It was triggered by this video from Lee Morris at the YouTube channel FStoppers. He titled his video ‘Google Just Made Photography Obsolete’, and for once, I do not think that this is hyperbole. His premise is this: the bread and butter of many professional photographers nowadays is what’s called product photography. These are the images you see featured on billboards, or in glossy magazine spreads, online and in other places where it is important to show off a product in its best light. A great deal of effort is put into this kind of photography: top-notch equipment (cameras and lenses that cost millions of shillings each); the most skilled photographers (who are paid hundreds of thousands of shillings per session); and painstakingly skilled editing. The result is a strawberry whose sweetness you can taste off the page. A bottle of perfume you can almost smell. A car ready to roar off a billboard and speed off into the distance. Morris says that the ability to create amazing results using Nano Banana and other tools will mean that product marketers will not bother engaging the services of photographers any more. If typing a brief prompt into a computer will give you professional-level pictures, why would anyone bother with hiring photographers and space, paying for their skills, and then waiting for weeks for the results, which may in any event be unsatisfactory?

This is a dismaying, and worrying, time for photographers, especially those of us who rely solely, or mainly, on earnings from the skill. The death of photography is not new – one need only ask the scruffy gents who used to hang around City Hall Way waiting to take a photo of people ‘holding’ the top of KICC. Now, those photos are taken on people’s mobile phones, and these professionals, despite their investment in instant printers, are mostly out of work. Professional headshots as well are going the way of film cameras: why bother turning up to a studio, carrying multiple changes of outfit, and being manhandled by a rude studio technician for the sake of a set of over-edited photographs for use in passport and visa applications, and for placing in your professional profile? Now, all you need is an imperfect selfie from your phone, which you can upload into an AI tool, which will then cough up a perfect head shot in whichever outfit and whatever pose you think appropriate. And since, increasingly, you never need to actually print any of these out, studios and their operators are no longer of value.

So where does that leave us? That is the subject of part two of this essay.