I just heard this for the first time. I'm using it. From now on.
A synthograph is an AI simulacrum that mimics the appearance of a photograph. Because face it, "AI photograph" is a contradiction in terms, an irredeemable misnomer. A photograph is written by light. A synthograph is created by a computer.
In his article "AI 'Street Photography' Isn’t Photography: What We Lose by Simulating Experience," which is where I heard this, Kōdō Shimon says, in a footnote, "The term 'synthography' appears to have been independently coined in 2022–2023 by both [Elke] Reinhuber and [Steph] Ango, demonstrating convergent thinking about AI-generated imagery. There are more people who could claim it around that time. It's a simultaneous invention."
In the next footnote he adds, "Baudrillard would likely recognize this as the 'precession of simulacra' at work: AI photography is not just a copy of reality, but a representation without original. Simulacra."
Not just a copy of reality, but a representation without original. There ya go. Thus, it's an illustration without being an artistic creation "by" anyone (despite being selected/directed, arguably), and certainly not a photograph because it doesn't have any direct or specific antecedent in the world of visual appearances. In photographs, there is no such thing as a totally accurate record, but usually you can provisionally retrieve something true from the picture about what was in front of the camera when the picture was made. That is, when Weston photographed Pepper No. 30, I believe there was truly a pepper there. I am not so depraved that I have to doubt it. He didn't invent it, he recognized it.
So this continues the dilution and corruption of photography as a direct record of the pretext in front of the lens when the picture is made, which has been traceably present more or less since the beginning but has gotten worse and worse as people figure out how to subvert it. The corruption of the photograph-as-record took a huge jump upward with digital and Photoshop et alia, and now with AI imagery has reached what must be its nadir. Please.
So as synthographs get better and better as AI does, leaving fewer and fewer signifiers of its use by which viewers can detect it, how are we to know we're looking at something wholly and cravenly unreal? Well, it's really no different than it ever was: the human maker is ultimately the one who either lies or tells the truth. Not the medium itself.
"Lying" is a judgemental term when it comes to expression. Albert Bierstadt's landscapes are fantasies; so are Oscar Reijlander's composites. El Greco's human figures are distortions; so are the funhouse-mirror pictures of Andre Kertesz. Monet's Rouen Cathedral series are impressions of light and color; so are the many experiments with motion blur by Ernst Haas. Edvard Munch was preoccupied with symbolism; so was Ralph Eugene Meatyard. Expression isn't lying.
But only because it isn't hiding. Yes, we'll be lied to by synthographs—probably a lot—and the "slap of truth" that honest photographs once possessed in more abundance will become yet more fugitive. Already we're at a point where the first reaction to any remarkable photograph is not wonder but skepticism. In an ideal world, we'd get to the point that a synthograph would be labeled as such, and the word "photograph" will be reserved for images that retain a connection to the real. Here's hoping.
I'll be using the term synthograph from now on, where and when it's warranted.
Mike
Original contents copyright 2024 by Michael C. Johnston and/or the bylined author. All Rights Reserved. Links in this post may be to our affiliates; sales through affiliate links may benefit this site. As an Amazon Associate I earn from qualifying purchases. (To see all the comments, click on the "Comments" link below or on the title of this post.)
Featured Comments from:
Very insightful discussion. My takeaway is that synthographs are to be accepted by the art world, not feared and loathed. And the creators (humans all) are the liars if they choose to be.
Posted by: Steve Deutsch | Tuesday, 24 December 2024 at 06:21 PM
But who owns the copyright, the AI or the individual telling the AI what it should create? And is create the proper term?
Posted by: JH | Wednesday, 25 December 2024 at 04:10 PM
At what stage does your photograph taken with a camera that has a lookup table to rectify / fix the lens vignette (one example) parameters (before it gets to the raw file) is more synthograph than photograph? The same question could be asked of Adobe / Dxomark with AI 'enhanced' noise reduction or scene calibration during file processing? Is there a line where a file becomes more synth than photo purely because your work flow incorporates significant computational enhancements or because the lens you've chosen is not fully optically optimised and needs a bios file to work?
Does that mean a mirror less camera is more a synth recorder than say a Leica rangefinder is a light recorder? :)
Seasons Greetings
Posted by: Robbie Corrigan | Wednesday, 25 December 2024 at 05:44 PM
Thanks for the introduction to that term, Mike. I really don't like that this is what we've come to in the realm of photography. I wish I held out hope that your "ideal world" would manifest itself. Alas, I don't.
Posted by: Mike Potter | Wednesday, 25 December 2024 at 08:29 PM
Baudrillard (and to a lesser degree Focault and Barthes) are much undervalued. They provide an accurate description of the postmodern world we live, reason and feel in, as exemplified in the article you cited. Can’t recommend reading them enough, if possible in the original french language. Their treatment of language as a malleable tool is admirable.
Posted by: Christoph Hammann | Thursday, 26 December 2024 at 12:52 AM
Recently I keep coming across this distinction about the results of AI: that they avoid the process. Uncritical AI advocates only care about the finished result, ignoring the importance of the process.
e.g. AI street photography - the importance of a person being there, noticing a unique scene, capturing the moment, the viewer knowing they're sharing a special scene that happened.
Or AI "paintings" or similar art - the importance of an artist making thousands and thousands of decisions during the making, of spending hours, days or weeks on perfecting the final result, the viewer knowing they're seeing the result of not only all the time put into making it, but the years the artist spent on improving their craft.
Or students using AI to write essays - the final essay in itself isn't the important thing, despite it being the thing that gets marked. The important thing is the process of researching, thinking, synthesising and writing. Learning to think and communicate. Skipping that step just to get a grade skips the learning that is the purpose of the task.
Posted by: Phil | Thursday, 26 December 2024 at 03:56 AM
Using that definition, most of the Tonalist landscape painters of the late 1800s to early 1900s were producing synthographic art. Painters like George Inness and J. Francis Murphy spent years drawing/sketching and painting outdoors to create a mental warehouse of imagery and visual tropes. Then they worked from memory and intuition, creating moody and emotionally resonant paintings of entirely imaginary scenes that nonetheless felt more "real" and meaningful than the photographs being produced in ever increasing numbers. It's well documented that many of the later Hudson River School painters like Frederick Church and Thomas Moran used photography as a tool for their giant studio paintings. Most of the Tonalist painters did not, creating images from their mental storehouse instead.
Posted by: Geoff Wittig | Friday, 27 December 2024 at 02:25 PM
Synthographs?
I am just getting used to Pixelographs.
Posted by: Daniel | Sunday, 29 December 2024 at 06:53 AM
'"AI photograph" is a contradiction in terms, an irredeemable misnomer.'
Can a misnomer contradict a misnomer? Do two misnomers combine into a worse misnomer or do they cancel each other out and yield a legitimate term? What if we "go with the flow" and simply concatenate this monstrosity to "artificial photograph"?
I admit, "synth____" might be useful coinage for a panoply of things now mislabeled "AI" this or that. Personally, I prefer "simul____" and the modifier "simulated", as in "simulographs" made by "simulated intelligence", as they more accurately capture the illusory nature of "AI".
On the other hand, taking Geoff Wittig's comment into consideration (though he's hanging an awful lot on temporality in his example), maybe we should try to emphasize the machine, as opposed to the synthetic, aspect; for example something like "computography" (though that sounds so 1980s).
Anyway, this will all be academic if my whimsy that the term "ersatz" will re-emerge to rescue us from this artificial hell actually happens.
Posted by: robert e | Monday, 30 December 2024 at 01:27 PM