Can you name them? Versions? I had a computer completely crash, and I lost everything, including my public-domain copy of Photoshop v. 2.0. (?) I have no interest in updates, because that free version did everything I needed it to. I’m left with using Paint and PowerPoint for graphic design.
No, I can’t, sorry. I gave up Photoshop ages ago (over 15 years, rather 20) when I switched from Apple to Microsoft. I went the road of GIMP first, then I bought the Affinity Suite, which is powerful enough for me, plus ClipStudioPaint for comics. They both offer life long licenses: the product you buy is yours. And it they are not prohibitively expensive, actually they often offer discounts and are then really cheap (IMO - talking about under 400$ for both programs). And GIMP, of course, is free.
Firefly is Photoshop’s new AI generative tool which is included in its latest Beta. It’s really quite impressive, even at this early stage of development. It does inpainting and outpainting and is lightyears ahead of Content Aware Fill and Neural Filters. Basically, you make selections, then tell Firefly what you want, and it does it. It gives you 3 variations at a time, and you can keep generating and refining via prompts, then save any variation you want. It’s very fast and accurate. It matches the tone, balance, shadows, lighting, and context of the photo precisely and convincingly.
One of its outpainting abilities is expanding the borders of photos any distance you want and Firefly will fill the borders to match the original photo so that you can’t tell where the extended borders begin. It even includes content-appropriate details, some of which are funny, but you choose the variation (s) you want to keep. You can, for example, take a photo in 16:9 aspect ratio and turn it into 9:16 easily. I’ve even used this for a couple of client photos to post on social media. Quite impressive.
Its inpainting is equally impressive. You can change the clothes of people, add or subtract jewelry, objects, animals, and people…whatever you want. The better you get with your selections and prompts, the better results you will get in return. It will even match the shadows and lighting that fall on (or from) the new objects you generate. It would take many hours to do this with cut and paste, and there’s no way you could do it as convincingly as Firefly. Firefly does it in seconds.
At this early stage of development, it’s not perfect. There is a bit of an uncanny valley with some of its human and animal generations, but with enough variations, you can find at least one that looks great. I look forward to future upgrades when they perfect the AI.
Here’s a good tutorial on Firefly by PIXimperfect (my favorite teacher).
Firefly is a game-changer for photo manipulation. It’s a tool I plan to use regularly for both small detail work, and large enhancements.
I’ve been playing with it for a couple days now, and I don’t find most of the results quite that convincing, though in some cases they are dead on. It’s extremely hit and miss for me. For example, yesterday I wanted to add a button onto the sleeve of a sport coat (button popped off) and it simply would not give me anything usable. It looked like a cutout of a button dropped on top of the photo.
So I just want to temper expectations. It is remarkable what it can do when it does it well, and this is really just the beginning, so I’m guessing in two years my complaints will be far away in my rear view mirror.
Beta apps are expected to have bugs, and Firefly is no exception. But, for it to start this strong is pretty remarkable.
Adobe does have a width limit for generations of ~1200 pixels (over that limit it becomes pixelated), but you can get around that by generating multiple smaller sections, then combining the layers. I expect the size limit to be lifted when it’s out of Beta…fingers crossed.
And, I find the better I get with prompts and modifying/fine-tuning selections, the sooner I get the results I want. There is a learning curve in order to get good.
I also get more “against our rules” errors than I should, for prompts that are in no way indecent. I expect that will be resolved soon, too.
All in all, color me impressed…but don’t use Firefly to make me look like a clown!
Heh. I just got one of those and I didn’t even prompt the thing! 
I wanted to recreate the bottoms of a toddler’s shoes, and I noticed it only gave me two options instead of three. I looked up and I got a “could not create because it goes against user guidelines”-type of message. I didn’t even suggest what to fill it with! What the hell did it try to generate? Unfortunately, I’m not having luck with getting some convincing bottoms of the shoe and content-aware is doing a better job for this particular task. I wish generative fill could look at the existing underlying layer and realize I want to try to keep the original pattern of the treads as best as it can. When I try to replace the whole sole, I get really, really odd textures.
Those error messages are annoying!
What Firefly can’t do (yet) is take instructions for photo enhancement/manipulation, it can only generate new graphic images where you tell it to via your selections (with matching tone, etc.) And, while it does reference images from its stock photo library and online, it can’t reference your photos (yet). But, it does do a fine job of analyzing the photo you’re working on when it renders generations (this is done in Adobe’s cloud).
As an experiment, I built a photo from the ground up on a blank canvas. (It’s for a client who sells boxed gifts and gift baskets online).
First I generated 3 people of mixed race sitting on a couch: one giving a boxed gift to the center person, and the other giving a gift basket to the center person. Then, I generated the background (inside a home library). Then I changed some of the clothes and colors on the people (to match the client’s logo colors). Then I added particular items into the basket (fruit, candy, a teddy bear, and flowers). Then for humor, I added a cat jumping onto the boxed gift.
Then I turfed the photo into neural filters to change the eye direction of one character and gave another a look of surprise.
I had to generate a lot of variations to get the look I was after, but all done within 90 minutes. It looks almost photo-realistic. I could make it entirely photorealistic if I spent a few hours with Photoshop’s conventional tools, but it looks good enough to post on social media right now.
And, just for fun, and to get my brother’s goat, I scanned an old B&W family photo from the early 1960s, when I was ~5. We were all seated at our old kitchen table in our old fashion kitchen. First, I expanded the borders on all sides. Then I generated us seated at a shiny new table in a modern kitchen (all B &W, with matching lighting, shadows, etc.). I put a black leather jacket on 85yo Nanna, with her motorcycle parked to one side of the kitchen, put a cop cap on Mom, and changed her clothes to something modern, a clown hat on sis with a polka dot dress, a bowtie and baseball cap on pop-pop, a crazy hairdo with flowers and a frilly shirt on big bro (lol), and cool shades and a snazzy outfit on me (of course).
I placed bottles of wine, whiskey and cigars on the table, along with shot glasses—all properly reflected on the table. I have each of us holding a glass of wine, or beer (even 5 yo me). Nanna’s holding a bottle of whiskey. And added a cat and a dog, similar to the ones we really had. Fun photo. Too fun…I gotta get back to work!
I believe this will soon be stiff competition for stock photo sites (including Adobe’s own). Why buy a stock photo when you can build one to your own specs?
We adopted a cat from the SPCA. The cat had been in a terrible accident (got caught in a fan blade when sleeping on a warm engine), and had broken shoulders and his skin knitted back in a way that pulls his eyes open. Everyone said he looks like a little alien:
This is Ty the cat:
We had just lost our Border Collie Katie, who was the dog of my kid’s childhood. To cheer him up, I made a photoshop story about how Ty was actually an alien, and his story of coming to Earth:
Ty started out playing in an alien band:
They were a bad influence, leading to Ty’s ‘lost years’:
He finally got away from them, and decided to start over on another planet.
Here is a rare photo of Ty arriving on Earth:
And here’s Ty today, all ready for new adventures:
Little known fact: the original concept for The Wizard of Oz did not test well with audiences.
Very well done, @Sam_Stone and @Tibby ! I am certainly in awe! You guys are quite skilled.
Sam, I cannot even guess at how you got that cat in all of those poses, and especially in that space helmet.
Tibby, your rework of the Wizard of Oz cast is truly warped–but in a good way. The Wicked Witch of the West in a polka-dot minidress? The Good Witch of the North showing off her legs? Munchkins drinking beer? The Tin Man with breasts? Warped, but hilarious!
Well done, both of you. Take a bow!
Thanks, @Spoons! I did that one with a little AI and a lot of Photoshop tooling. And, yes, most of my for-fun stuff tends toward warped (blame Mad Magazine for warping me long ago).
Would love to see more of your, @Sam_Stone, and other’s photo manipulations. This is a fun thread, worth keeping alive.
Don’t know whether this counts as manipulation, I think linking it does no harm.
Come to think of it, there are so many levels of manipulation, metamanipulation and manipulators in this picture that it is almost funny on a level I did not intend. I call it Rocket Man and Racket Man.
@Tibby, thank you. I read Mad Magazine when I was younger (hell, let’s face it, if it was still publishing, I still would), and I know what being warped means. At least, according to Mad.
I have plenty more manipulations, and I’d love to show them off, but I am not in your or Sam’s class as far as Photoshop goes. But what about something that doesn’t use Photoshop, or computer design in any way? What might you think about this:
That was totally by hand, by me. It may not fit in this thread, but I am very proud of it, and the more I get to show it off, the better.
I’m a calligrapher at heart, really, and I always like to show my work off. I’ll get back to photo manipulation in my next message or two. Meanwhile, pour yourself a dry martini and enjoy!
That’s quite good, and effective—makes me thirsty for a martini, even at 9-am.
I used to order my martinis with 3 olives: 1 pimento, 1 almond, and 1 blue cheese stuffed. I ordered them in a big martini glass to leave enough room for the gin.
I like calligraphy, though I know little about designing with it.
That’s funny!
Here’s a caricatured variation on the theme.
Hey, it’s five o’clock somewhere. 
It’s like any other piece of art, really—you don’t just sit down and knock off something worthy of display in a gallery in an hour. You plan and prepare, you do a number of studies to see what works and what doesn’t. You practice with your tools, especially if they are unfamiliar to you–I am very comfortable with any of the Speedball C-series pen nibs, but I’d want to practice with any of the Speedball B-series nibs before actually using them.
“Gin” required the use of a ruling pen on a compass-type setup. I had never used one before, so I practiced a lot. That work (“Gin”) took about a week for the planning and execution, and most of that was planning and practice—the actual execution took about four hours.
Thing is, that with calligraphy, once you do something, it’s done. Unlike Photoshop and similar tools, you can’t just “Undo” and try something else. So planning and practice is essential before executing what will be a finished work worthy of display.
It was fun. I haven’t done it for a while; maybe I should get back into it.
Oh, I am a typography nerd myself, and have been designing a font for the last 25 years or so (will be finished any day now, for sure). Calligraphy is also a hobby of mine, but the last years I have been doing it on a computer screen with a stylus. It is not as neat as real ink, but you can correct your mistakes. Only I don’t use photoshop, I disagree radically with their licensing policy, I use Clip Studio Paint, which is quite good enough for me. And the Affinity Suite and GIMP, but way more seldom.
@Tibby: Glad you liked it! Those are the kind of things I have been doing to practise for my real project: a comic for which I would like to use the font mentioned above that will also be finished any year now. If you follow the pictures chronologically as they are displayed I believe that the pics got better with time. Technically, at least.
Next time you’re in Florida, be sure to take your kids to The Tragic Kingdom in Dismal World.
Does this count?
It’s an iPhone shot- my homage to Feininger,
Maybe I’m missing something, but I’m not sure what I’m looking at. It’s a small-town street, under a cloudy sky. And who or what is “Feininger”?
Little help?