Can You Have Photography without AI?
Can You Have Photography without AI?
A badge of honor, perhaps, but why?
First Thing
Thank you for your help getting word of this newsletter out! A bunch of you are new subscribers, and I appreciate that you’re here. And thanks to those of you who sent feedback about what you’d like to see in future editions. Keep ’em coming.
AI Backlash Undercurrent
As part of the cruft that accumulates in my Facebook feed are posts from some photo-related groups I’ve joined over the years. I usually skim past them, quickly taking in the submitted photos and skipping the camera models for sale and the like. I should really go trim those sources, but…it’s easier not to? That’s kind of the problem of being on Facebook in general, which of course is what Facebook counts on to keep you checking in and creating impressions. But I digress.
An uncharacteristic short post stopped my scrolling for a moment. “None of my photos have ever been touched by AI!” said one anonymous photographer. (I’m paraphrasing.) That part of me that reads and responds to comments wanted to jump in with a hearty laugh and refute the claim, but I’ve learned to squelch that impulse because I don’t need to wind myself up like that.
And yet, the comment lingers with me, for two reasons:
1. I can think of many reasons why that person is most certainly wrong. But to give him the benefit of the doubt, it’s entirely possible he shoots and develops only film, or he’s talking about a specific set of curated photos that he sees as being his best shots. Or maybe he’s referring to generative AI apps and services versus camera-captured photos.
But it’s more likely he does have a digital camera, which right there is increasingly a tiny box full of AI. And if it’s a smartphone, that’s mostly AI. (But he’s probably not counting smartphone photos.)
He could instead be referring to editing his photos. Some apps are AI-forward: Luminar Neo is built on an AI engine (the predecessor was called “Luminar AI,” even), and many of the newest features and marketing bullet points for other image editors such as Pixelmator Pro and ON1 Photo RAW rely on machine language to perform tasks. But let’s again give him the benefit of the doubt and guess that perhaps he uses Lightroom, but hasn’t touched any of the new ML-assisted masking features.
Has he ever clicked Lightroom’s Auto button? Since at least 2017 the output from clicking Auto has been based on Adobe Sensei (the blanket term for Adobe’s ML technologies). Has he ever run facial recognition to identify people in his library? Or maybe he edits with Capture One, which is powerful and popular and is just now wetting its toes in the AI pool with its Smart Adjustments feature.
My point is that even if he’s never touched his photos using AI features, AI features have touched his images in some way.
(As you can imagine, this is why I didn’t jump into the discussion. Invariably it would devolve into an art-versus-machine rant that I just don’t have time to rabbit-hole right now.)
2. AND YET. The other reason the comment stuck with me is because it was made in the first place. AI is mentioned even more frequently in the news lately, so it’s on a lot of photographers’ minds. But not too long ago saying “None of my photos have been touched by AI!” would be like exclaiming, “No one needs automatic transmission when this clutch pedal works perfectly fine!” (There probably are people who still say this. And a lot of younger folks wondering what the hell I just typed.)
Just as we no longer say “digital camera” unless it’s in the context of film cameras, we soon won’t make a distinction about whether AI is involved in our photography. It simply will be, in large and small ways. And in most cases, AI will make image capture or editing easier or less time-consuming. AI, for all kinds of reasons that fall under “better” and “worse” is the present and future of photography.
Add AI to... Manual Focus?
This is the type of headline that twists my brain: It’s Time to Add Some AI to Manual Focus. Here’s a Way to Do It (DPReview).
Camera manufacturers are finally getting more serious about incorporating machine learning into their camera bodies, predominantly in the autofocus systems. The newest cameras can lock onto a bird’s eyes, for example. The interplay between the camera sensor, image processor, and lens motors in milliseconds is frankly impressive.
So what’s this about manual focus? Manual focus lenses break that entire chain.
However, Matt Waller makes an interesting point: employ AI to help with MF Assist features such as focus peaking or enable more precise manual focusing. He writes:
Yes, AI MF Assist. Let manual focusing have a share of the camera’s subject-recognition smarts. Have the Peaking outline appear only on eyes, only on birds, only on…er, motorbikes. (Hey, motorbikes move, that actually would be very helpful as you try to follow them from behind the fence.)
With the AI smarts behind it, the peaking outline would remain on the detected subject as they hop through the branches or feint through the defenders. It would allow a skilled camera operator to keep even a moving subject in focus without losing composition.
It’s a neat idea, and opens possibilities for camera companies to leverage their technologies competitively.
Drone Detection Mode?
Speaking of focusing, Fujifilm has just released a firmware update for the Fujifilm X-H2S that adds autofocus object recognition for insects and…drones. The camera already includes modes for birds and airplanes. I can see how insect recognition would be helpful for nature and macro photographers, but who other than perhaps YouTube reviewers have been scrambling to focus on drones?
This sounds to me like some engineers figured it wouldn’t take much more work to add the shape of common drones to the Airplane dataset. Hey, why not?
Sometimes Do Read the Comments
I started this post with a comment, so it feels right to end with a comment (despite the adage that you should never read the comments in an article, which is mostly accurate). Last week I linked to DPReview’s article about Adobe using customers’ images to train its AI, and this comment made me laugh out loud:
themountainphotographer Well, if Adobe use my photos that will degrade the AI.
That has been my secret plan to screw Adobe all along.
Let’s Talk
Thanks again for reading and recommending Photo AI to others who would be interested. Send me any questions, tips, or suggestions for what you’d like to see covered at jeff@jeffcarlson.com.