To access Generative Fill, first download the Photoshop BETA app in the Creatve Cloud installer. Next, turn on "Synthetic Fill" in the preferences under "Early Access". To turn on the contextual task bar, enable it in the "Window" menu under "Application Frame" and "Tools".
I got it working. Installing the Beta doesnt install the latest version. So I then had to go to the updates tab click "check for update" then install the new update. I still didnt have any options to select, but generative fill was automatically installed and working then anyway
The amount of times I heard Aaron issue a nervous existential chuckle throughout this video encapsulates what I feel about this, what a crazy place we're at right now
I felt the same way as Aaron's laugh sounds when I heard the announcements today while in the middle of making a tutorial! It is totally mind-blowing...
We can't trust photographic evidence of ANYTHING anymore. These are so good! You've blown my mind about what can be done now with a few loose strokes and a descriptive vocabulary. I love your teaching style -- clear, concise and quick.
I was thinking the same thing - but I wonder if film photography will be a way of bringing 'truth' back to photography. Can you imagine being required to do film photography to prove authenticity!
@@MrSindriSvan But that would require to see only prints - as soon scan the film to show online your are back to questions of authenticity. This is why many photo competitions now require you to have the raw file available to prove originality.
I just spent several hours editing new sod that was laid down for a wedding venue. The sod had brown edges, where it was joined together, it was a mess. If I had waited one day for this update, I probably would’ve had six hours extra with which to edit their portfolio. SMH. Sheesh. I knew when Photoshop announced this, that I could come straight to your channel to learn how to do it. I am amazed, and don’t short yourself, because your talent and the way you teach makes things like this so simple to understand and comprehend.. thankyou Sensei Aaron!
This is nuts! I want to now go back through my whole catalog and see what I can play with to remove things I didnt like in the past but knew would take WAY too long to do it. This is absolutely insane and I can't wait to see what the demand will bring.
I am blown away by the new stuff Photoshop has created, but the best part of this video is hearing Aaron's expression when he sees what is generated. "WHAT!". This makes me want to try it even more now. You have been my teacher from the beginning and I have a learning disability. You made everything fun and easy to learn. Thank you for always continuing to teach us new things. Thank you! ♥
That is the great thing about choices. You get to decide if you want it or not. I will enjoy it and might pay for it too. Who knows. :) @@WilliamCruzing4photos
Generative fill only produces imagery up to 1024x1024 pixels. So, if an image is expanded, the whole new area generated by the AI it will be MUCH lower resolution than the original artwork. The workaround is to generate in small sections, this keeps the quality high.
That's the same way people are avoiding the size limitations of Dall-E 2 and other AI generators by adding small chunks to slowly build large images. Emphasis on slow. I've read that the size limitation of Generative Fill is 2000x2000 pixels, not 1024x1024, which makes it the largest currently on the AI photo market, I believe. To test, I took an image of mine that was about 1900x1900 pixels into the Beta and it worked fine. Then I refilled some areas with no problem and then expanded it in incremental slices with Generative fill to about 3500 x 2500 pixels. The resolution seems relatively consistent. More testing required.
Colin Smith at Photoshop Cafe just posted a PS action to automatically do generative fill in multiple 1024x1024 sections to get around the resolution limitation.
Having painstakingly removed objects, changed backgrounds, added more pleasant sky layers, extended space above and below the original captures many many times for my self and others this tool is easily bests my efforts in almost ever case. Very impressive!
A lot of designers and photographers will be threatened by the power of AI - but I'm SUPER stoked about these tools since it makes the mundane aspects of design faster so we can get closer to creating whatever we can dream up. We're in uncharted and very exciting times.
Bro you’re like a kid in a candy shop…20+ year photoshop user here…flexed this over the weekend…blew me away it actually looks at layers and all sorts of inputs…nuts…and the non destructive preservation…woahhhh….this is sick and if this is V1 of this…the future is fkn bright…slam dunk 🏀
Heavens above. This feels like when content aware fill just dropped but 100x better. As someone that uses content aware fill and similar tools daily this really is beyond a gamechanger. Wild, Adobe.
I kinda like that it makes the annoying work simpler its going to allow me to completely focus on the creative aspects instead of wasting hours on the dumbest little details my creativity is much better when prompted by outside forces so it will even aid in my own creativity, very exciting
I've been really impressed with the remove tool. However, this morning I decided to give generative fill a try, and I must say, I see incredible potential in it. As a travel photographer, I often capture moments like a young couple riding a bike down a charming cobbled street in Seville. But when I review the photos later, I often notice distractions like an ATM machine, cars, or a movie sign that take away from the original scene. In the past, it would take me hours of painstaking work using tools like clone stamp, healing, and layers to simplify the image and recreate the atmosphere I experienced when taking the shot. But now, with generative fill, I can achieve even better results in just 10 minutes or less. It's a significant improvement from my early days spent in a dark room, manipulating images through burning and dodging techniques.
by removing those elements you're not recreating the atmosphere you originally experienced, you're fabricating a sanitised fake version of it. Would you like classic old photos from the 1920s to have the lovely old cars, movie signs, etc to be airbrushed out?
@@jogennotsuki He won't need to travel any more, he can just get AI to do it from his armchair. He can get AI to not frame the shots properly so he can then use more AI to remove the ATM machine, cars, or a movie signs, that added to the scene
The information you provide is always clear and easy to understand, but your enthusiasm, after all these years, is - to use your term - mind blowing. Many thanks!
I'm now a professional photographer in in my journey of learning the editing process you really helped me along the way and talked me a lot. And first and foremost I want to thank you for that because I wouldn't be the person I am today without giving you credit. I am now on my own TH-cam channel and starting my own platform. A lot of the tools that were needed for me to create my own style we're found online and you were a big part of that. So thank you once again.
A large part of the effectiveness of the generative fill feature is how you craft the prompt - it actually uses the linguistic descriptions to inform its generation. For example in the video you asked it to generate "clouds" which produced an unrealistic rendering of a cloud. However by tuning your prompt (e.g. "a clear blue sky with white clouds") you would get a much better result as it has more information to generate a comprehensible replacement as a new layer.
An absolute game-changer and we don’t even know it’s full capability yet. I’ve been using photoshop since I was 17 and it’s truly a wonder to see how far it’s come.
This is insane!!!! It’s great in a way because I have to remove things from images all the time, and I also make composites for work so this will reduce my work flow immensely. However it is showing that it’s one step closer to making this possible for anyone to do so I’m predicting at some point being a photoshop editor as a job will be a thing of the past.
If you look in close at 15:16 with the removed cars, it has left in some tire and car body because the selection didn't overlap the parking meter. It's somewhat reassuring that some human input and understanding of the tool is still required, at least at this stage.
I'm blown away. It took me all these years to be able to refine my skills to the point where people would offer me money for it. There's literally no skill needed to do this stuff with this tool. The only thing people will need to learn how to do is how to navigate UIs and command the software to do what you want. But since it's not perfect, it's still good to have some knowledge with editing. While I enjoy the process of editing, doing it this way can turn an hours long project into seconds. Big time saver!
yeah, we've have reached a point where Ai does a better job than most professional editors, give it 2-3 years and it will be PERFECTION, the problem is, nobody is going to pay you to remove something when Ai can do it faster, better, and still create more variations. Ps will be accessible to more people who have zero understanding of how compositing works.
13:20 that is not out of focus/depth of field done by A.I. that is because the generated areas are max 1024 pixels on the longest side. This is something worth keeping in mind. You will loose definition if you generate areas wider or taller than 1024 pixels. A work around for this is to generate in patches.
I've been learning from Aaron for over ten years, and he's made me a master at Photoshop. So while Adobe Generative AI/Firefly will bump everyone to a pseudo-advanced level regarding Photoshop proficiency, his students will take that advanced new level to exciting new heights. Without that expertise, others will create images that aren't artful, seem somewhat imbalanced, or seem AI-generated. Thank you, Aaron, for this exciting artistic journey. You should make a Pro class on fixing the real image issues after the Generative AI quick fixes to create stunning and highly credible issues. It would be a big seller.
It's superb and great for re-editing backgrounds and images for video too. For architectural, great for removing cars, bins, people etc. Just saves time.
I've noticed the limitations of the matching perspective is that it doesn't add in atmospheric perspective. The cloud for instance could use a 10% blue layer over the top to make it look like it is in the distance. But it is fantastic tool!
I started horsing around with this yesterday and it's truly amazing. One thing it can't do -- I was looking at a pic of some kids playing frisbee on a beach, outlined an area in the ocean and asked for a shark fin. I got dozens of whale flukes. :-) Seriously though, it's going to create huge workload for people-- all those images you thought were junk and worthless? Guess what-- you have to go back through them and think about how generative fill can change them-- there's probably some keepers in there now!
I had a large photo scanned from the 1940s of a large oil well. It was vertical and I was exporting to 16x9 for video. There was a barn in the photo, 80% to 90% was cropped out. I highlighted all the blank space to the left of it, hit fill with ZERO prompts and it not only recreated the barn but the grain matched, the color (fade browns) matched, the glow and film quality matched, it created trees.. it honestly blew my mind. Client would have thought it was the original photo if I hadn't of told him.
I'm an Aaron as well I've used photoshop probably about 22 years now, and I've seen so much progress in that time with Photoshop. You never stop learning. Some of these features are absolutely mind blowing, but I think from a professional standpoint they're going to be a massive game killer for professional designers who maybe self employed to potentially lose money... There was a time when I use to color and restore old photo's, but nowadays with the AI technology out there, people without any element of skill can have their photo restored and colored in a couple of clicks. To be fair unless you're targeting old people with no computer tech knowledge. A friend of mine who is also a designer has lost a lot of revenue already because of it with some of these AI apps and tools, that people without skill can create almost the results they want. Don't get me wrong these features are very cool, but what it will mean is that the future of new Photoshop users won't be as skilled at Photoshop editing as you or I in the future. I think these features could be a real killer for a lot of self employed designers or even small companies.
At first, I was like "Well, there goes any chance of making photography a source of income", but as you said, this cuts so much time off of editing if you are already making money. I do like what you mentioned what Adobe will do to protect, or keep the copyright to it's source. And yes, this is addictive. I am now going through some of my old shots, and throwing some "pop" in there. I know there are some "purists" who will think this is cheating, but I dont think it is as photography is subjective. It is still "your" art, your photograph. I guess the end state though is.....what is real, vs imagined? Great video as always :-)
When the last photo removed the cars and generated an entire pier sidewalk, I had to stop the video and walk away. THIS IS INSANE. Literally hours, and sometimes even days, of work done in 10 seconds to perfection. These are the kinds of features that somewhat justify Adobe's ridiculous subscription payment model. GAME CHANGER!
Your last point about justifying the sub is absolutely spot on. For non-pros like me, only ever interested in creating really nice but amateur works, this is a serious point. We amateurs will often have some skills - like talent with posing our subjects, with choosing our subjects, with creating interesting images - but we lack the whole skill set for doing all the really hard stuff like Aaron goes through in this video. This would give us the freedom we‘ve never had.
Agree completely. Every time I get the feedback request from Adobe I say the same thing - great product, the negative is cost. Because of the subscription model it is very expensive. This certainly takes some of the sting away.
If anything like Stable diffusion then the prompt you type in is quite important. If you don't get exactly what you want work on changing your prompts to be more specific. i.e. instead of [clouds] use [fluffy white dispersed clouds] etc.
Playing with this right now. if you have on location model shots that have been just taking up space, grab them and go to town.... works EXACT as shown in the video. LOVE it
As someone who has been learning photoshop for the last 10 years using videos like these, this makes me feel pretty uneasy. It took real skill and an artists eye to composite things into a photo and look photorealistic. Now that can be done in one click and it’s hard to come to terms with the lack of artistry that goes into it now.
You’ll still need that skill and artistic eye, it’s still valuable. For those that haven’t mastered it, even with using this tool, photos will look over done to some degree.
Sorry, but when I was first introduced to Photoshops demo of them removing a scratch from a car in the early 1990's, it blew my mind. Now nothing does. Maybe you are scared that this is available to unskilled people and your job security...
@@mabelandbea let's be realistic here. AI is only gonna get better and I would challenge everyone in this comment section to compete against it for free. Nobody is gonna pay you when Ai hmdoes it better. Faster. And if you don't like the result. Just generate another variation with one click 😅
“Real skill” is a delusion based on the time you were born into, people born tomorrow would have a different perspective based on that as well. Adapt, evolve.
A game changer for sure, I have mixed feeling about this. The technology is astounding, however, as a student of Phlearn, Aaron taught a lot of us how to create a composite image. The right way, with perspectives, and the lines going out toward the horizon, the videos on Phlearn are insanely detailed and amazing. learning new techniques you'd build up a skill set. Something, not everyone can do. However, now anyone with no skill or knowledge of compositing can turn out amazing composite images over and over with little more than words. The technology is great, but will it do to graphic designers, what the streaming did to video stores? If someone with no skill can turn out amazing images, what will become of graphic designers, or those who taught a lot of us Photoshop?
They will adapt to the changing tech like everyone else does. If you fight the future you get left behind. Artist should have the skills they have built up to stand out among the AI prompt only style of "user". Now combined really amazing artist with large set of skills with this AI tools and bam we will start even more amazing content. The average users content will always be average even all dolled up with AI.
That's a fair concern, but photo editing and compositing are just a few small parts under the umbrella of graphic design. For certain design projects where data/content is the main focus, I'm not sure how much AI will be able to help in terms of designing layouts that make sense for a human viewer.
Just remember where photography began... It has, and always will be, married to the technological advancements that gave rise to it and continue to progress it forward into the future. In a hundred years something will come along that makes all this look like flash powder...and that's how it should be.
@@Slojurq Spot on. And I expect that when photography was invented the doomsayers were saying it would be the death of painting. There's nothing new about these reactions.
I am simply trying to change the color of a white sweatshirt to pink. I gave generative fill a hex color and it came back with some crazy results... it made the sweatshirt into a blue hoodie, a polo shirt and a blouse. Have a suggestion on how to simply change from white to pink, while keeping shadows?
I was playing around with this a bit today - it’s certainly fun, but I didn’t find the results consistently great - tried replacing subjects clothing with different colours, replacing subjects etc - but found pretty mixed results, and it’s all limited by being low res for now too. I do like the idea of being able to describe what you want to do in plain English rather than having to master menus etc - that may actually be the most transformative change to the PS interface if it’s applied more broadly
Crazy how excited PS creators are over this feature. Never seen so many videos over the SAME feature at one time. And every presenter is freaking giddy! Wild.
This is amazing and a game changer for everyone. I love your videos and tutorials and am enjoying my subscription so far. Thanks for such amazing content!!!!❤
I produce mainly video content. I often need to get b-roll for my videos that are old photographs from people I've interviewed. Often you get a photo that's been shot in portrait orientation, and you wish that it'd had been taken in landscape to match the format of your video. I've already used the generative fill a bunch of times to extend out old photos. I used to have to do this by stretching pixels out and hoping that it looked okay, or by cloning and hand painting pixels. But this is so much faster and looks pretty much seamless.
That last image when you removed the cars and people was quite surreal, but also having a pink ginger as a cloud - that's a new way of looking at it haha. I wonder where this advancement will stop.
I have the bar but the whole Generative fill section is missing - no idea where to turn on "Synthetic Fill" cant find any preferences under "Early Access".
Problem is I apparently didn't provide my date of birth way back when I first set up my Adobe account. Now I am unable to use this feature (and presumably others) because my account shows I am "not 18 years old yet". Er..., I'm in my 50's. And there appears to be NO WAY to update my profile. Any suggestions?
Everything but the credit system sounds amazing to me haha... But working in tech and in AI i know firsthand how much processing power is involved in running AIs... This technology is available open source and for FREE elsewhere but as also someone who has been using Photoshop for over 20 years this will be nice to have built into Ps... Very cool...
There is both a Photoshop and a Photopea plugin for the open source software you've mentioned. Not quite as easy to use, but a good alternative in case the credit system is egregious.
@@NorthgateLP i have seen a few pop up and AI inpainting has been a feature around for awhile in Stable Diffusion but i played with the Ps beta today. It is pretty amazing to have it create masked layers directly in Photoshop... But just as when Dalle2 went from beta to a credit system i indulged for a short while then never looked back after I was charged for every blank error and/or poor generation... I felt with a credit system it is hard to set up when results will always vary... Even some of my tests today came out amazing and other use-cases were absolutely horrible... But it's AI and still really powerful yet quick evolving technology... I guess I will see how my sponser responds to the credit system... Or if Creative Cloud Subscribers will get a a daily or monthly amount... I guess we will find out in September when the next Photoshop releases... As someone who gets 90% of his income from Adobe Software I think AI will make most of Adobe's products obsolete in the next few years if they dont fully integrate the tech in various ways... 2 years ago I was overwhelmed with people asking me how to learn photoshop and after effects etc... In 2023 everyone just asks me how to use AI... I cant predict the future but i think most realists see the writing on the wall... Haaha... but im glad Adobe is at least trying to intergrate this technology in an interesting way...
Good point. After using Stable Diffusion a lot and now playing around with this feature, it will all depend on the price. Sometimes it will take a lot of repeats until the AI produces the result you want and being charged for every generation can become a money consuming disadvantage. Another point is the rules for generation. I tried to switch a bikini into a vintage swimsuit, but always ran into the warning-message, that this violates the rules. Is this a tool for photographers or a kindergarten-tool for puritans? That's far too restrictive for a professional photographer.
@@tvncgn that is true too... Censorship will play a big role as well... Even the newest Stable Diffusion Models all have strict NSFW filters so i had to train my own models if i wanted blood instead of ketchup for example... another part about Firefly/Adobe AI is no settings controls... The simplicity can make it more of a toy and less of a professional feature... I used Dalle2 before Stable Diffusion and when that went to a credit system i never used it again after i was being charged for bad generations and errors... I felt they needed some sort of leeway system... I felt it was like walking into a restaurant ordering a pepsi and the waiter giving me an orange juice then forcing me to drink it since I paid in advance... Maybe a bad metaphor but i dont like paying for services that may not always provide what i want... hhaha... Every week a new AI app or service is popping up too so I feel by the time Adobe figures it out there will be so many more options that it will not even matter... but either way great to see progress... I felt if Adobe didnt get on the AI train then theyd be taking the bankruptcy bus at some point haha... Just from personal experience as someone who was teaching Adobe Softwares on the side... Two years ago everyone came to me asking about Photoshop or After Effects etc. now they all just want me to teach them AI... I mean it took me years and lots of practice, studying and doing to get where I am... Now someone can type one sentence and get amazing results without even knowing how to spell the word photoshop... The future of creativity will be complicated... hahaha... I'm just trying to stay positive and hope for the best...
I just tried to remove some objects in a very complex streetphoto with a lot of reflections and stuff... like magic! Even at 200% I can't notice any errors. Scary! Just for fun, I generated a biker and a mother with her child in the background. Looks 100% realistic, including shadows and light.
I knew Phil way back when he was just Clone Phil. He showed great promise as he grew up, becoming known as Content Aware Phil. Now, he's fully mature as Generative Phil.
I cannot get the contextual bar to open without selecting something first, how did he bring it up to select subject or remove background? Only options after it opens via selecting something is generative fill. Thanks in advance
Just for fun I thought I'd try adding a more abstract background to a photo I took and had it generate shiny spheres behind the model. Not only did it work, it generated a reflection of THE BACK OF THE MODEL on the spheres 🤯🤯🤯
Just wow. I’m so glad I’m those kind of people who doesn’t like to delete images. I can’t wait until I go back and see what I can do with my old pictures.
Amazing video, thank you sooo much! Do you know how I can use the same Gen fill background for multiple pictures? I'm trying to change the bg for my son's graduation pics ugly backdrop.
i am a bit frustrated. i have been trying to get this option but it is just not appearing. i have the contextual task bar without Generative fill. i updated both Photoshop and Photoshop beta and i am using an 8 bit image in RGB color mode but still no luck
I've been trying to access this for a while now and can't seem to get it to show. I have the most recent beta, I'm under Photoshop preferences, down to Early Access - Beta, but all it says is "see beta features at menu -> window -> beta feedback", but there's no way to toggle this on like in the video. Anyone else able to get past this?
To access Generative Fill, first download the Photoshop BETA app in the Creatve Cloud installer. Next, turn on "Synthetic Fill" in the preferences under "Early Access". To turn on the contextual task bar, enable it in the "Window" menu under "Application Frame" and "Tools".
Minimum PC Configuration??
Any ideas why some of us have no options showing under the early access tab?
I got it working. Installing the Beta doesnt install the latest version. So I then had to go to the updates tab click "check for update" then install the new update. I still didnt have any options to select, but generative fill was automatically installed and working then anyway
I have no option for synthetic fill in the beta early access area??
@@spiritnovel Follow the steps that I posted above and it should fix the issue.
The amount of times I heard Aaron issue a nervous existential chuckle throughout this video encapsulates what I feel about this, what a crazy place we're at right now
Exactly, his nervous laughter says it all... ALL his skills and countless hours... "puff, vanish". Terrifying
I felt the same way as Aaron's laugh sounds when I heard the announcements today while in the middle of making a tutorial! It is totally mind-blowing...
And all the 'what!?'s 😂
I remember a few years back I learnt from you how to cut out a subject using the pen tool manually, and now we're here. It's truly mindblowing!
So much progress!
He taught me how to create an entire background. Bless you guys
We can't trust photographic evidence of ANYTHING anymore. These are so good! You've blown my mind about what can be done now with a few loose strokes and a descriptive vocabulary. I love your teaching style -- clear, concise and quick.
And now blockchain technology for authenticating photo 📷
I was thinking the same thing - but I wonder if film photography will be a way of bringing 'truth' back to photography. Can you imagine being required to do film photography to prove authenticity!
Photoshop has been around for about a quarter of a century, this is nothing new.
@@MrSindriSvan But that would require to see only prints - as soon scan the film to show online your are back to questions of authenticity. This is why many photo competitions now require you to have the raw file available to prove originality.
I just spent several hours editing new sod that was laid down for a wedding venue. The sod had brown edges, where it was joined together, it was a mess. If I had waited one day for this update, I probably would’ve had six hours extra with which to edit their portfolio. SMH. Sheesh. I knew when Photoshop announced this, that I could come straight to your channel to learn how to do it. I am amazed, and don’t short yourself, because your talent and the way you teach makes things like this so simple to understand and comprehend.. thankyou Sensei Aaron!
This is nuts! I want to now go back through my whole catalog and see what I can play with to remove things I didnt like in the past but knew would take WAY too long to do it. This is absolutely insane and I can't wait to see what the demand will bring.
I am blown away by the new stuff Photoshop has created, but the best part of this video is hearing Aaron's expression when he sees what is generated. "WHAT!". This makes me want to try it even more now. You have been my teacher from the beginning and I have a learning disability. You made everything fun and easy to learn. Thank you for always continuing to teach us new things. Thank you! ♥
thank you so much! your words mean a lot to Aaron and the whole team!
Don't worry, enjoy the free aspect of the tool, won't belong before they add a massive price to it. Lol
That is the great thing about choices. You get to decide if you want it or not. I will enjoy it and might pay for it too. Who knows. :) @@WilliamCruzing4photos
Generative fill only produces imagery up to 1024x1024 pixels. So, if an image is expanded, the whole new area generated by the AI it will be MUCH lower resolution than the original artwork. The workaround is to generate in small sections, this keeps the quality high.
That's the same way people are avoiding the size limitations of Dall-E 2 and other AI generators by adding small chunks to slowly build large images. Emphasis on slow. I've read that the size limitation of Generative Fill is 2000x2000 pixels, not 1024x1024, which makes it the largest currently on the AI photo market, I believe.
To test, I took an image of mine that was about 1900x1900 pixels into the Beta and it worked fine. Then I refilled some areas with no problem and then expanded it in incremental slices with Generative fill to about 3500 x 2500 pixels. The resolution seems relatively consistent. More testing required.
Colin Smith at Photoshop Cafe just posted a PS action to automatically do generative fill in multiple 1024x1024 sections to get around the resolution limitation.
Having painstakingly removed objects, changed backgrounds, added more pleasant sky layers, extended space above and below the original captures many many times for my self and others this tool is easily bests my efforts in almost ever case. Very impressive!
A lot of designers and photographers will be threatened by the power of AI - but I'm SUPER stoked about these tools since it makes the mundane aspects of design faster so we can get closer to creating whatever we can dream up. We're in uncharted and very exciting times.
Bro you’re like a kid in a candy shop…20+ year photoshop user here…flexed this over the weekend…blew me away it actually looks at layers and all sorts of inputs…nuts…and the non destructive preservation…woahhhh….this is sick and if this is V1 of this…the future is fkn bright…slam dunk 🏀
Love your enthusiasm. I've been a photographer for 48 years. What a game changer!
Heavens above. This feels like when content aware fill just dropped but 100x better. As someone that uses content aware fill and similar tools daily this really is beyond a gamechanger. Wild, Adobe.
Yeah. This is on a whole other level. It analyzes the area and preserves the same angle, lightning etc
I kinda like that it makes the annoying work simpler
its going to allow me to completely focus on the creative aspects
instead of wasting hours on the dumbest little details
my creativity is much better when prompted by outside forces
so it will even aid in my own creativity, very exciting
I've been really impressed with the remove tool. However, this morning I decided to give generative fill a try, and I must say, I see incredible potential in it. As a travel photographer, I often capture moments like a young couple riding a bike down a charming cobbled street in Seville. But when I review the photos later, I often notice distractions like an ATM machine, cars, or a movie sign that take away from the original scene. In the past, it would take me hours of painstaking work using tools like clone stamp, healing, and layers to simplify the image and recreate the atmosphere I experienced when taking the shot. But now, with generative fill, I can achieve even better results in just 10 minutes or less. It's a significant improvement from my early days spent in a dark room, manipulating images through burning and dodging techniques.
by removing those elements you're not recreating the atmosphere you originally experienced, you're fabricating a sanitised fake version of it. Would you like classic old photos from the 1920s to have the lovely old cars, movie signs, etc to be airbrushed out?
How is it feel living through a groundbreaking technology era. Do you feel lost sometime, sir?
So what you're saying is that you're a travel photographer who can't frame shots properly...
@@jogennotsuki He won't need to travel any more, he can just get AI to do it from his armchair. He can get AI to not frame the shots properly so he can then use more AI to remove the ATM machine, cars, or a movie signs, that added to the scene
The information you provide is always clear and easy to understand, but your enthusiasm, after all these years, is - to use your term - mind blowing. Many thanks!
I'm now a professional photographer in in my journey of learning the editing process you really helped me along the way and talked me a lot. And first and foremost I want to thank you for that because I wouldn't be the person I am today without giving you credit. I am now on my own TH-cam channel and starting my own platform. A lot of the tools that were needed for me to create my own style we're found online and you were a big part of that. So thank you once again.
A large part of the effectiveness of the generative fill feature is how you craft the prompt - it actually uses the linguistic descriptions to inform its generation. For example in the video you asked it to generate "clouds" which produced an unrealistic rendering of a cloud. However by tuning your prompt (e.g. "a clear blue sky with white clouds") you would get a much better result as it has more information to generate a comprehensible replacement as a new layer.
An absolute game-changer and we don’t even know it’s full capability yet. I’ve been using photoshop since I was 17 and it’s truly a wonder to see how far it’s come.
This is insane!!!! It’s great in a way because I have to remove things from images all the time, and I also make composites for work so this will reduce my work flow immensely. However it is showing that it’s one step closer to making this possible for anyone to do so I’m predicting at some point being a photoshop editor as a job will be a thing of the past.
100% correct !
If you look in close at 15:16 with the removed cars, it has left in some tire and car body because the selection didn't overlap the parking meter. It's somewhat reassuring that some human input and understanding of the tool is still required, at least at this stage.
It is insane how quickly things are changing. This is mindblowing!
I'm blown away. It took me all these years to be able to refine my skills to the point where people would offer me money for it. There's literally no skill needed to do this stuff with this tool. The only thing people will need to learn how to do is how to navigate UIs and command the software to do what you want. But since it's not perfect, it's still good to have some knowledge with editing. While I enjoy the process of editing, doing it this way can turn an hours long project into seconds. Big time saver!
yeah, we've have reached a point where Ai does a better job than most professional editors, give it 2-3 years and it will be PERFECTION, the problem is, nobody is going to pay you to remove something when Ai can do it faster, better, and still create more variations. Ps will be accessible to more people who have zero understanding of how compositing works.
13:20 that is not out of focus/depth of field done by A.I. that is because the generated areas are max 1024 pixels on the longest side. This is something worth keeping in mind. You will loose definition if you generate areas wider or taller than 1024 pixels. A work around for this is to generate in patches.
Yes, it is depth of field done by AI. Just because its low res doesn't mean it will look out of focus.
I've been learning from Aaron for over ten years, and he's made me a master at Photoshop. So while Adobe Generative AI/Firefly will bump everyone to a pseudo-advanced level regarding Photoshop proficiency, his students will take that advanced new level to exciting new heights. Without that expertise, others will create images that aren't artful, seem somewhat imbalanced, or seem AI-generated. Thank you, Aaron, for this exciting artistic journey. You should make a Pro class on fixing the real image issues after the Generative AI quick fixes to create stunning and highly credible issues. It would be a big seller.
It's superb and great for re-editing backgrounds and images for video too. For architectural, great for removing cars, bins, people etc. Just saves time.
This is nuts. I started with 3.0 in the 90's. I can only imagine what Photoshop will be like in 10 years.
In 10 years Ai will be crushing every single artist combined. It's crazy
I'm blown away. Omg. This saves me hours and hours of my time. Thank you so much for sharing this.
I've noticed the limitations of the matching perspective is that it doesn't add in atmospheric perspective. The cloud for instance could use a 10% blue layer over the top to make it look like it is in the distance. But it is fantastic tool!
I have been seeing videos about this all day and I was waiting to see how excited YOU were about this lol
Explaining sooo clear and calm.. enjoy every time watching
We love it! Making awesome art has never been so accessible.
I just can’t believe what I’m seeing! I’m blown away! 🤯
This is so impressive. Bravo Adobe for giving us this magical resource.
I started horsing around with this yesterday and it's truly amazing. One thing it can't do -- I was looking at a pic of some kids playing frisbee on a beach, outlined an area in the ocean and asked for a shark fin. I got dozens of whale flukes. :-) Seriously though, it's going to create huge workload for people-- all those images you thought were junk and worthless? Guess what-- you have to go back through them and think about how generative fill can change them-- there's probably some keepers in there now!
I had a large photo scanned from the 1940s of a large oil well. It was vertical and I was exporting to 16x9 for video. There was a barn in the photo, 80% to 90% was cropped out. I highlighted all the blank space to the left of it, hit fill with ZERO prompts and it not only recreated the barn but the grain matched, the color (fade browns) matched, the glow and film quality matched, it created trees.. it honestly blew my mind. Client would have thought it was the original photo if I hadn't of told him.
I'm an Aaron as well I've used photoshop probably about 22 years now, and I've seen so much progress in that time with Photoshop. You never stop learning.
Some of these features are absolutely mind blowing, but I think from a professional standpoint they're going to be a massive game killer for professional designers who maybe self employed to potentially lose money...
There was a time when I use to color and restore old photo's, but nowadays with the AI technology out there, people without any element of skill can have their photo restored and colored in a couple of clicks. To be fair unless you're targeting old people with no computer tech knowledge. A friend of mine who is also a designer has lost a lot of revenue already because of it with some of these AI apps and tools, that people without skill can create almost the results they want.
Don't get me wrong these features are very cool, but what it will mean is that the future of new Photoshop users won't be as skilled at Photoshop editing as you or I in the future.
I think these features could be a real killer for a lot of self employed designers or even small companies.
Thank you so much for sharing your thoughts with us and we absolutely get where you're coming from!
At first, I was like "Well, there goes any chance of making photography a source of income", but as you said, this cuts so much time off of editing if you are already making money. I do like what you mentioned what Adobe will do to protect, or keep the copyright to it's source. And yes, this is addictive. I am now going through some of my old shots, and throwing some "pop" in there. I know there are some "purists" who will think this is cheating, but I dont think it is as photography is subjective. It is still "your" art, your photograph. I guess the end state though is.....what is real, vs imagined? Great video as always :-)
When the last photo removed the cars and generated an entire pier sidewalk, I had to stop the video and walk away. THIS IS INSANE. Literally hours, and sometimes even days, of work done in 10 seconds to perfection. These are the kinds of features that somewhat justify Adobe's ridiculous subscription payment model. GAME CHANGER!
Your last point about justifying the sub is absolutely spot on. For non-pros like me, only ever interested in creating really nice but amateur works, this is a serious point. We amateurs will often have some skills - like talent with posing our subjects, with choosing our subjects, with creating interesting images - but we lack the whole skill set for doing all the really hard stuff like Aaron goes through in this video. This would give us the freedom we‘ve never had.
Agree completely. Every time I get the feedback request from Adobe I say the same thing - great product, the negative is cost. Because of the subscription model it is very expensive. This certainly takes some of the sting away.
And we are into AI territory and deception. Bad Karma
Don’t worry… they’ll be raising the prices again soon. 😆
Great video and demo of this amazing tool Aaron. I can think of so many photos I've taken in the past where this will prove super useful. Cheers
If anything like Stable diffusion then the prompt you type in is quite important. If you don't get exactly what you want work on changing your prompts to be more specific. i.e. instead of [clouds] use [fluffy white dispersed clouds] etc.
This is either liberating or terrifying. I haven't decided which. Thanks for your breakdown. Very intriguing!
Playing with this right now. if you have on location model shots that have been just taking up space, grab them and go to town.... works EXACT as shown in the video. LOVE it
How have you been? It's great to see you! Thanks again for all your expertise.
As someone who has been learning photoshop for the last 10 years using videos like these, this makes me feel pretty uneasy. It took real skill and an artists eye to composite things into a photo and look photorealistic. Now that can be done in one click and it’s hard to come to terms with the lack of artistry that goes into it now.
well wellcome to 2023
You’ll still need that skill and artistic eye, it’s still valuable. For those that haven’t mastered it, even with using this tool, photos will look over done to some degree.
Sorry, but when I was first introduced to Photoshops demo of them removing a scratch from a car in the early 1990's, it blew my mind. Now nothing does.
Maybe you are scared that this is available to unskilled people and your job security...
@@mabelandbea let's be realistic here. AI is only gonna get better and I would challenge everyone in this comment section to compete against it for free. Nobody is gonna pay you when Ai hmdoes it better. Faster. And if you don't like the result. Just generate another variation with one click 😅
“Real skill” is a delusion based on the time you were born into, people born tomorrow would have a different perspective based on that as well.
Adapt, evolve.
wow first time in a while a tutorial that actually worked thanks man
Thanks for watching!
A game changer for sure, I have mixed feeling about this. The technology is astounding, however, as a student of Phlearn, Aaron taught a lot of us how to create a composite image. The right way, with perspectives, and the lines going out toward the horizon, the videos on Phlearn are insanely detailed and amazing. learning new techniques you'd build up a skill set. Something, not everyone can do. However, now anyone with no skill or knowledge of compositing can turn out amazing composite images over and over with little more than words. The technology is great, but will it do to graphic designers, what the streaming did to video stores? If someone with no skill can turn out amazing images, what will become of graphic designers, or those who taught a lot of us Photoshop?
They will adapt to the changing tech like everyone else does. If you fight the future you get left behind. Artist should have the skills they have built up to stand out among the AI prompt only style of "user". Now combined really amazing artist with large set of skills with this AI tools and bam we will start even more amazing content. The average users content will always be average even all dolled up with AI.
Its just a tool to be used like any other
That's a fair concern, but photo editing and compositing are just a few small parts under the umbrella of graphic design. For certain design projects where data/content is the main focus, I'm not sure how much AI will be able to help in terms of designing layouts that make sense for a human viewer.
Just remember where photography began...
It has, and always will be, married to the technological advancements that gave rise to it and continue to progress it forward into the future. In a hundred years something will come along that makes all this look like flash powder...and that's how it should be.
@@Slojurq Spot on. And I expect that when photography was invented the doomsayers were saying it would be the death of painting. There's nothing new about these reactions.
I have been playing with this. It’s amazing. It’s not perfect but when it hits it’s amazing.
WOW WOW WOW, you blew my mind with this. Thanks!
I am simply trying to change the color of a white sweatshirt to pink. I gave generative fill a hex color and it came back with some crazy results... it made the sweatshirt into a blue hoodie, a polo shirt and a blouse. Have a suggestion on how to simply change from white to pink, while keeping shadows?
I was playing around with this a bit today - it’s certainly fun, but I didn’t find the results consistently great - tried replacing subjects clothing with different colours, replacing subjects etc - but found pretty mixed results, and it’s all limited by being low res for now too. I do like the idea of being able to describe what you want to do in plain English rather than having to master menus etc - that may actually be the most transformative change to the PS interface if it’s applied more broadly
Your videos are fantastic!! This new tech is MIND BLOWING!!
They may come close, but nothing can beat you, Aaron Nace
Crazy how excited PS creators are over this feature. Never seen so many videos over the SAME feature at one time. And every presenter is freaking giddy! Wild.
This is amazing and a game changer for everyone. I love your videos and tutorials and am enjoying my subscription so far. Thanks for such amazing content!!!!❤
Yep, I knew they would be charging extra for this feature. Thanks for the heads up.
This is going to help many of us who could not edit before, in the other hand, many editors simple task jobs might be lost over time.
Everybody will he on the same level. It's the same as everybody using cheats on call of duty.
I produce mainly video content. I often need to get b-roll for my videos that are old photographs from people I've interviewed. Often you get a photo that's been shot in portrait orientation, and you wish that it'd had been taken in landscape to match the format of your video. I've already used the generative fill a bunch of times to extend out old photos. I used to have to do this by stretching pixels out and hoping that it looked okay, or by cloning and hand painting pixels. But this is so much faster and looks pretty much seamless.
Yes, Gamechanging!
Now we are even more reliant on cloud services and their subscriptions. Really nice
Great video! btw I drive past that cafe all the time!
16:40 So would something like that cloud error cost a credit?
Scared and amazed at the same time. Wow! That's exciting!
That last image when you removed the cars and people was quite surreal, but also having a pink ginger as a cloud - that's a new way of looking at it haha. I wonder where this advancement will stop.
I'm falling in love with this feature. Awesome
It is so good!
Congratulations on being natural and so informative.
I have the bar but the whole Generative fill section is missing - no idea where to turn on "Synthetic Fill" cant find any preferences under "Early Access".
Thank you Aaron.
I replaced a whole picture: ctrl+A then generate fill
Try it and enjoy 😅
Oh my god 😮. This is absolutely awesome. You are the best Aaron! Thanks a lot!
Way talented photoshop editor ❤
Problem is I apparently didn't provide my date of birth way back when I first set up my Adobe account. Now I am unable to use this feature (and presumably others) because my account shows I am "not 18 years old yet".
Er..., I'm in my 50's.
And there appears to be NO WAY to update my profile.
Any suggestions?
Thanks sir for this valuable information..🙏🙏
completely speechless with this new stuff
Haven't clicked a Phlearn in a couple years! Glad I'm back
We're glad to have you back!
Everything but the credit system sounds amazing to me haha... But working in tech and in AI i know firsthand how much processing power is involved in running AIs... This technology is available open source and for FREE elsewhere but as also someone who has been using Photoshop for over 20 years this will be nice to have built into Ps... Very cool...
There is both a Photoshop and a Photopea plugin for the open source software you've mentioned. Not quite as easy to use, but a good alternative in case the credit system is egregious.
@@NorthgateLP i have seen a few pop up and AI inpainting has been a feature around for awhile in Stable Diffusion but i played with the Ps beta today. It is pretty amazing to have it create masked layers directly in Photoshop... But just as when Dalle2 went from beta to a credit system i indulged for a short while then never looked back after I was charged for every blank error and/or poor generation... I felt with a credit system it is hard to set up when results will always vary... Even some of my tests today came out amazing and other use-cases were absolutely horrible... But it's AI and still really powerful yet quick evolving technology... I guess I will see how my sponser responds to the credit system... Or if Creative Cloud Subscribers will get a a daily or monthly amount... I guess we will find out in September when the next Photoshop releases... As someone who gets 90% of his income from Adobe Software I think AI will make most of Adobe's products obsolete in the next few years if they dont fully integrate the tech in various ways... 2 years ago I was overwhelmed with people asking me how to learn photoshop and after effects etc... In 2023 everyone just asks me how to use AI... I cant predict the future but i think most realists see the writing on the wall... Haaha... but im glad Adobe is at least trying to intergrate this technology in an interesting way...
Good point. After using Stable Diffusion a lot and now playing around with this feature, it will all depend on the price.
Sometimes it will take a lot of repeats until the AI produces the result you want and being charged for every generation can become a money consuming disadvantage.
Another point is the rules for generation. I tried to switch a bikini into a vintage swimsuit, but always ran into the warning-message, that this violates the rules.
Is this a tool for photographers or a kindergarten-tool for puritans?
That's far too restrictive for a professional photographer.
@@tvncgn that is true too... Censorship will play a big role as well... Even the newest Stable Diffusion Models all have strict NSFW filters so i had to train my own models if i wanted blood instead of ketchup for example... another part about Firefly/Adobe AI is no settings controls... The simplicity can make it more of a toy and less of a professional feature... I used Dalle2 before Stable Diffusion and when that went to a credit system i never used it again after i was being charged for bad generations and errors... I felt they needed some sort of leeway system... I felt it was like walking into a restaurant ordering a pepsi and the waiter giving me an orange juice then forcing me to drink it since I paid in advance... Maybe a bad metaphor but i dont like paying for services that may not always provide what i want... hhaha... Every week a new AI app or service is popping up too so I feel by the time Adobe figures it out there will be so many more options that it will not even matter... but either way great to see progress... I felt if Adobe didnt get on the AI train then theyd be taking the bankruptcy bus at some point haha... Just from personal experience as someone who was teaching Adobe Softwares on the side... Two years ago everyone came to me asking about Photoshop or After Effects etc. now they all just want me to teach them AI... I mean it took me years and lots of practice, studying and doing to get where I am... Now someone can type one sentence and get amazing results without even knowing how to spell the word photoshop... The future of creativity will be complicated... hahaha... I'm just trying to stay positive and hope for the best...
Watching this video while trying everything in PS and it blows my mind!
All doable in stable diffusion. Can even use it as a plugin for your (fully purchased from adobe) version of photoshop. Offline.
I just tried to remove some objects in a very complex streetphoto with a lot of reflections and stuff... like magic! Even at 200% I can't notice any errors. Scary! Just for fun, I generated a biker and a mother with her child in the background. Looks 100% realistic, including shadows and light.
yeah, and this is just one year technology, give it 2-3 years and it will be pure perfection!
This is a game changer for portrait photographers. I've been having a little bit of a play with it and just wow!
I knew Phil way back when he was just Clone Phil. He showed great promise as he grew up, becoming known as Content Aware Phil. Now, he's fully mature as Generative Phil.
I cannot get the contextual bar to open without selecting something first, how did he bring it up to select subject or remove background? Only options after it opens via selecting something is generative fill. Thanks in advance
This is crazy! Amazing! Does it work for illustrations?
With the yarn art example, it's amazing how it even generates shadows to go along with the yarn art.
Good bye to interior designers. Anyone can just Photoshop photos of their own home and test what they like
Wow, this is mind-blowing! I luv it
Do you have a video on adding a fountain to a photo in PS or Lr classic
Great review!
Good lordy, the third example was mind-boggingly amazing.
Such a great tutorial!!
Love the photos of Pismo beach too❤
Just for fun I thought I'd try adding a more abstract background to a photo I took and had it generate shiny spheres behind the model. Not only did it work, it generated a reflection of THE BACK OF THE MODEL on the spheres 🤯🤯🤯
Just wow. I’m so glad I’m those kind of people who doesn’t like to delete images. I can’t wait until I go back and see what I can do with my old pictures.
Amazing video, thank you sooo much! Do you know how I can use the same Gen fill background for multiple pictures? I'm trying to change the bg for my son's graduation pics ugly backdrop.
not available yet on my photoshop beta
Click the Check for Updates button in CC. I didn't see the update until I did that. I think it was just added today.
@@Poukibear thank you very much ,i m updating now
And so it begins… it’ll take us quite a long time to catch up with what’s now possible - this is one of the cool things. 🤞🏼
Great video!!
Great video
You forgot the video chapter marker at 13:39 and 15:47 tho
i am a bit frustrated. i have been trying to get this option but it is just not appearing. i have the contextual task bar without Generative fill. i updated both Photoshop and Photoshop beta and i am using an 8 bit image in RGB color mode but still no luck
I have concerns about the credit system. We already have a credit system with our subscription to Adobe, which already has an ever-increasing cost.
Excellent video. Thank you!
Holy Moly! incredible stuff. Can't wait to test this out.
So if I use the lasso tool and select myself, and tell it to reduce my weight by 20 Lbs. would that make me look thinner?
I've been trying to access this for a while now and can't seem to get it to show. I have the most recent beta, I'm under Photoshop preferences, down to Early Access - Beta, but all it says is "see beta features at menu -> window -> beta feedback", but there's no way to toggle this on like in the video. Anyone else able to get past this?
Got the same issue your mentioned... It works anyway but maybe not at its full potential.