Can we talk about AI again?
I last talked about it back in issue 045 when I called AI a magic trick minus the magic. So if you read that you’ll know I’m basically a photography purist who thinks AI should keep its grubby little fingers out of music and art.
But since then Adobe decided to let AI stick its grubby little fingers into photography through the latest Photoshop Beta release and something called “Generative Fill”.
My first reaction to this was, WHAT THE <the generated text was removed because it violated user guidelines>, ADOBE?!
But I do tend to overreact to things. So I thought I should at least try it and spend some time using it before passing opinion through superchARJ.
I’ve just edited a wedding and it gave me my first chance to see what it can do and to work out how and why I might use it in my editing workflow.
So now I’m ready to share my balanced opinion with you.
Photoshop isn’t new.
We’ve always had, at our fingertips, the ability to do what this generative AI can do, if we had the willpower to learn how to do it.
Now it’s just much more accessible to those of us who’ve never bothered to learn how to properly retouch in Photoshop.
So it’s important to say that a lot of this isn’t new, Adobe have just taken all the skill and talent out of it.
And that’s what a lot of AI does in general, by the way. But I’ve had that rant before and will have it again, so let’s stay on topic.
But let’s be clear that we’ve always been able to, relatively easily, add and remove things from our photos. Now it’s just so easy a toddler could transform a street scene into a beach scene with a random gurgle and a couple of mouse clicks.
And no, I don’t buy that ‘prompting is a talent in itself’. Just no.
It already it feels like the only people who call stuff “AI” are the same people who still say the “WWW” when giving you the website address to something.
Generative Fill is what this new photoshop feature is called. And here’s how it works.
You make a rough selection, you click ‘generate’, you enter a prompt if you want to add something specific, or you leave it blank if you want photoshop to work it out for itself, and you click generate again.
If you want to add something specific, that’s what you put in the prompt box.
Here’s a photo I took recently.
They are real clients of mine so I disguised their identities simply by drawing a very rough lasso selection around each head and typing ‘disguise’ in the prompt box.
This was a suburban wedding at a hotel venue.
I decided to use Generative Fill and my imagination to come up with the image below. The only skill this took was for me to make selections, type very simple prompts, and click generate.
So first I added completely fake sides and top to the image like I’ve seen a lot of people doing in reels and tiktok demos of this feature. Then I replaced tarmac with a beach, added a boat, palm tree, some coconuts, a sandcastle, a cactus, a cat, some bits and pieces on the sand, an old boat and a skyscraper. And obviously the disguises again. Took less than 3 minutes. And that’s on my old, slow iMac. You M2 ultra users will be able to achieve this level of art in the blink of an eye!
I originally added a starry night sky too, but I took that off because who’s gonna believe that?! Gotta keep it believable.
Ok it’s ridiculous. An absolutely ridiculous thing to do to someone’s wedding photo in my opinion, but I think we’re going to see quite a bit of it, just like we saw a lot of spot colour when that first became fairly simple. And do you remember the phase of dinosaurs chasing bridal parties? I think that may make a return. Yippee.
If you don’t remember that from the first time round, I envy you. Buckle up.
As I say I just edited a real wedding and while I normally take zero percent of my photos into photoshop, I probably took 20 or so images into photoshop because this new tool is amazing at two things.
1. Removing distractions.
2. Adding a bit of extra space at the edges if your composition was off.
I love the masking tools within Lightroom and my prediction is that it won’t be long until this generative technology will make its way into a Lightroom update.
But for removing things like fire exit signs, annoying stuff on ceilings, fire extinguishers or anything that’s made its way into your photos, this generative fill tool is WAY better.
It’s so good and so easy. It’s not as quick as using Lr masks, but it’s better enough that I think for now I’m likely to do my cleaning up in Photoshop.
I had an image where I should’ve ideally stepped back a bit. There was a tiny bit of the bride’s train cut off and I’d have just preferred a little more breathing space around her.
I increased the canvas size by 10%, selected the new 10% and did a generative fill (no prompt).
It did a perfect job and even generated the extra little bit of dress. It also just means if the photo is ever framed there’s enough breathing space that the frame won’t cut important bits off.
I don’t want to be doing this often, but if necessary it works really well.
I predict short term overuse, followed by it just settling into the many tools we have available to us as 21st century wedding photographers. We’re going to see some really amusingly bad stuff being created, as well as some really amazing stuff being created. Amazing because we will have no idea it’s ‘generative’.
I do wonder if it’ll make us a little lazier now it’s just getting easier and easier to ‘fix in post’. I don’t know. I hope not. Like with the adding extra bits to photos. I was impressed by the technology, but I was annoyed at having to do it, just like I’m annoyed at myself when I have to remove stuff that I could’ve noticed and removed real-time with better compositional awareness. Having to generate new bits to photos in my latest edit has made me want to make sure I don’t make the same mistake in future.
But it’s great to know how good the tool is if I ever need it, that’s for sure.
I recommend you try it out.
Don’t knock it til you’ve tried it etcetera blah blah blah.
I will remain a relative photography purist and hope that my ‘heavy’ editing is restricted to dodging and burning but now I know how this works it opens little doors in your mind and I’ll definitely use it if I need it. Why wouldn’t you?
But I probably won’t transform any more inland suburban weddings into tropical beach weddings just because it’s easy to do so.
Weddings are real life, real time events.
I want to believe that my clients don’t want fake sides added to portrait images just because it’s possible.
I don’t think they want every single ‘distraction’ removed from photos just because that’s what photographers tell other photographers is important in critiques. What’s a distraction to you and me, might be something deeply important or sentimental to our clients.
I think of my job as making the raw, authentic, reality of the wedding look as good as possible (by my own personal definition of ‘good’) with my settings, vision, skills and a shutter click.
The editing process, for me, is where I correct for consistency and add my ‘polish’. It’s where my photos are finished, not where they’re created.
We should hang our hats as photographers on the talent and skill we have with the camera in our hands and to our eyes, not on the fakery that’s possible (and has always been possible) in Photoshop.
What do you think? Agree? Disagree? It’s all just food for thought.
Thanks for reading!
Adam
