We've been testing the fill feature and I have to say it's pretty amazing. We actually couldn't believe how accurate the fill is. We had a vertical image of two subjects in a train station (surrounded by trains on either side and wanted to see if it could fill for a horizontal. Well, it filled in the trains and walkway perfectly and even added correct colors/grading, lighting and DOF. We also tested it out with an image of a couple during fireworks on a balcony with hundreds of people below BUT there was an annoying table with drinks and a mans leg in the shot (couldn't ask him to move). The fill perfectly rebuilt the railing with spindles, filled in people below and added directional lighting. Amazing. I'm not a big fan of Adobe and their politics but on a purely technical level this is a huge positive development.
It's insane what this and AI generated artwork can do. And this is only just the start of it - It's still hit or miss sometimes, but in a year or two it'll be so much better, and it's already pretty amazing. What I love most about it (other than speeding edits up by a lot, and making previously impossible things now possible), is that it does allow so much more control over creativity. Things you couldn't do before or could only imagine in your head, are now much more achievable. Previously my skills might not have been good enough to get exactly what I wanted, or to fix things that really annoyed me about a photo, but this literally does it in seconds, and usually way better than I could have done it. Combining this with Midjourney means it's a pretty exciting time for creative work - but it does also mean you can no longer trust any image you see as being real.
I don't think I'd call it creative work per se. It's more like uncreative work. The AI software is doing the "creative" part. Instant gratification (especially when used by non-creatives), with AI being the workhorse part of the art or photography, will just subdue real human creativity by laziness and lack of pursuing skills and fresh, new ideas and concepts.
Photography is now more like painting than ever before. Creating the changes is not "fake / false" but merely allowing photographers to become creative painters. This is beyond impressive and opens massive improvement potential for any "creator." Thanks for your video.
@ANASTASI LINE does painting only become painting when you get dirty? What if you use paint while staying clean? And what if your mouse is so dirty that you get messy while using the computer? Hmm. Besides, there are many paintings known by well known painters which are "painted over". So that is no argument.
Integrating this with Photoshop makes the whole graphic and visual design scene even more crazier . Let see where this takes us . Results are really astonishing
Amazing. I'm so excited to play with it more - I've already used the ai generative fill to expand photos and it's truly mind-blowing. Would love to see more videos for inspiration.
I love it. Our design team does as well. We’re always asked for extracurricular things to spice up images in post, or remove very difficult things like cranes, forklifts, woodpiles and tricky license plates. This has greatly sped up mock-ups and workflows. It’s light years better than content aware algorithm and this is still in its infancy, let’s see how accurate it gets 2-5 years from now.
I spent some time learning this new feature in Photoshop yesterday. I love it and am excited to see how it will evolve. This is another excellent tool I will use when needed, and understand it is limited so I will always have to put my finishing touches on the final image. I had some things I wanted to accomplish with Midjourney and could not with its deep learning curve. This PSD AI feature is a walk in the park. Of course, there will be those who will exploit it and think they are now real artists by using it by creating tacky art, just like the millions who believe they are real photographers using their cell phones & DSLRs on auto mode.
Great experimentation here, Glyn. I love how it handles the shadows and lighting from original image. Also a great way of adjusting to somehing that seems obvious is here tostay. It truly is the creative mind that produces the final image and AI is just another, albeit phenomenal tool.
You can tell how advanced of a leap it was because the first reaction from everyone was to get scared... like we'd just seen fire for the first time and it looked like magic.
While the assertion you mentioned emphasizes the role of our minds, hands, and tools in expressing creativity, there are indeed other perspectives to consider. Here are a few alternative viewpoints: Environment and Inspiration: Creativity can also be influenced by external factors such as the environment we inhabit and the experiences we encounter. Our surroundings, interactions with others, and exposure to different cultures and ideas can all contribute to our creative expression. Emotion and Intuition: Creativity often emerges from a deeper level of emotional and intuitive understanding. It is not solely a product of our cognitive processes or the tools we use. Our feelings, instincts, and personal experiences can play a significant role in shaping our creative output. Collaboration and Interactions: Creativity can be enhanced through collaboration and interaction with others. Engaging in discussions, brainstorming sessions, and receiving feedback from peers can foster new ideas and perspectives, transforming individual creativity into collective creativity. Constraints and Limitations: Sometimes, limitations and constraints can actually fuel creativity. When faced with challenges or restricted resources, individuals may find innovative ways to overcome these obstacles and generate unique solutions or artistic expressions. Imagination and Vision: Creativity often originates from the power of imagination and the ability to envision possibilities beyond the present. It involves the capacity to think beyond what currently exists, to dream, and to visualize novel concepts or artistic creations. Inner Exploration: Self-reflection, introspection, and personal growth can also contribute to creativity. Exploring our own thoughts, beliefs, and emotions can lead to self-expression and the generation of original ideas. It's important to note that creativity is a complex and multifaceted concept, and different individuals may have unique perspectives on what fuels their creative process. The interplay between our minds, hands, tools, environment, emotions, and interactions ultimately shapes how we express our creativity to the world. Generated by ChatGPT - but I pretty much agree :-)
I love it. Gives you complete creative freedom to express yourself in ways that are now less limited. I don't have huge amounts of money for a studio, so using AI can really plug that gap for me.
Brilliant video! You're spot on with your comment at the end. I've had ideas floating around in my head for years that I just don't have the time or energy to produce due to having to learn new several new software packages extensively. Now I can take my main skills of PS compositing and color grading and combine them with Ai to produce these art pieces. It's about using the tools that we are giving to our advantage.
I haven't had a chance to try the Generative Fill yet but Unmesh Dinda (Piximperfect) states that the fill resolution is limited to a cap of 1024 pixels on the long side which implies that you need to keep you "fill" selection areas within this "cap" to maintain the resolution of bigger images. The examples you've shown don't seem to address this Glyn - so I'd be interested in your view on this issue. In the wider context, whilst I agree that creative images makers will use these tools to add to their creations - my concern is that Ai technology will be another big step in driving photography and design down to the lowest common denominator, with the result that it will become even more difficult for professional photographers and designers to make a living. As a book cover stock photographer it concerns me that the market will eventually be inundated with Ai generated images and designs that have never, as it were, seen the inside of a camera but generated from text prompts alone. There has already been one high-profile example of this where Bloomsbury used an Ai generated image on the cover of "House of Earth and Blood".
When I was a production artist I can think of spendiing two days retouching a photo to clean up a beach to be used for a big boating campaign. With this generate function, I am thinking less than an hour would have been needed. This is powerful for things like that. For that this is beyond good. For the creative mind this is exponential .
I think it is great. Although it blurs the line of what is real and what is not ... soon enough we won't be able to tell the difference. I wonder what effect that will have in how we perceive our reality and how we process it. Video is next ...
this is insane, I have been playing with it with mixed results, the good results are really crazy good, the bad results are comically bad, but of course its in Beta, once it has been improved and understood with user submited data it is going to be insane
Wow, you've got some great use cases there. At this point it's just a tool to augment our own creativity. It's great for cleaning up images, and obviously for so much more. I'm looking forward to seeing how it progresses, but in my mind it will never be the end of photography. A large part of photography is just being out, camera in hand, and I can't replicate that with AI.
@@glyndewis Yeah but no one will care if it's taken outside or generated. AI is here to replace the craft, anyone that thinks otherwise is delusional. Just give it time and we'll all be replaced.
@@glyndewis Clicking and clicking again and again is also not creative in editing. It´s a useful tool for layouting, when the photo is cropped too tight, but I cant see any creativity in turning portrait to landscape.
I agree with the last part for someone who are creative and professional its an added tool, for others will be just filling their portfolio with unrealistic photos, and people by their nature will not care to distinguish between authentic photos that uses the tool to enhance or just completely generated by AI. The value and story behind a photograph will not matter as AI eliminates the hassle of going out and shoot your photos ( in majority of none photographer’s mind set)
Initially, I started to feel threatened by AI fearing that i would be out of work very soon. But then i decided I'm still the one in charge of creativity, and I'm just going to make AI my bitch, and consider it another powerful tool in the box.
I've been experimenting with it to change or extend backgrounds; I've made a blank wall slightly less blank by adding some faded graffiti to it; and, most egregiously, I livened up a boring corner by materializing a cat in it (Generative Fill did an excellent job on the cat.) But here's the kicker: This technology depends on the machine-learning model being "trained" on huge sets of prototypical images, which it slices and dices to execute your generative prompt. So the more your concept is commonplace, mundane, and obvious, the more likely the MLM is to have been trained on something similar, enabling it to produce excellent results. Meanwhile, the more idiosyncratic, unique, and distinctive your original concept, the less likely the MLM will find similar source materials in its training set, and the more likely it is to produce an "off"-looking result. You can almost use it like a naffness meter -- the more it improves your image, the weaker your original idea was in the first place...
Hi Glyn yet another great video. Can you tell me if the Remove tool uses pixels from the original image and not add pixels it’s got from else where. I do know you don’t need the internet to use it so it’s not using the cloud to generate the pixels for the image. I’m just researching the ethics of the AI generative tools for competition purposes. Thanks
Need a new computer to access features. I ordered it yesterday and will take 2 weeks to arrive. Hopefully Photoshop does not become obsolete before then, and humans manage survive Ai development
I'm glad you did this video. It's helped me do better photography. I'm now going through all my photos and deleting every one of them where I borrowed unneeded help from other photos. I want to be REAL.
In the portrait of the farmer, the generated areas of rock & grass lack what is important in those areas of the original, the fine gritty texture. Simply adding more stuff that resembles the original portion says that the artist didn't understand what was of value, ie what made the look valuable. In fact were the final version the original, it would have been improved by cropping back to the vertical version. So while this is a great tool it does nothing for visual judgement.
@@yeohi But what about those, who you seem to have ingnored, who "manipulate" their photos ever-so-slightly, well short of 'illustration'? Is it not a game-changer for them too?
It's worth noting that Generative fill only does 1024x1024 pixels and stretches the result. So large backgrounds can be grainy and artifacty - although there is a workaround which is easily internet findable. I've found this handy for shots where I haven't quite left enough background and need more space in the photo. I'm finding AI a bit of a giant everythingburger - if you don't know what you're trying to achieve you'll end up with everything, and kind of, nothing...
You've made by far the most compelling video on Adobe AI Generative Fill, taking all the really important perspectives into account. No drama, just advantages with a philosophical ending. Very very nicely done, thanks.
I just used the remove tool to remove a blurry robin from in front of my bird photo of a juvenile starling. using the remove tool. I then used generative fill to generate a new tail for my bird. After a second try, I got a near perfect result!
I've always used alternative software to Photoshop, but honestly this is a game changer and I'm considering switching. Are there any requirements to use this tool? Also, do you need to pay extra to enable it, or is it enabled with an ordinary Photoshop subscription?... And, is there a feature to add objects to a photo? I've recently started making framed wall art mockups using stock photos of interior rooms, but it's so time consuming to make the frame blend in with the room and look realistic. If I could just write a prompt to add a blank frame on a wall, and if it looks as realistic as the chair's legs, that would save me like 99% of the time I spend doing it manualy
I love the creative opportunities and the ability to enhance one's own original work. The main thing that concerns me though (with AI generation generally, not just Adobe's Generative Fill) is where the replacement / additional elements are being harvested from other photographers' work. For instance, were all the hair options for the viking created from scratch by Photoshop? Far more likely they were derived at least in part from the work of another unknown, unpaid photographer. Next time it may be your work making-up someone else's photo. Is that saleable as an original work? Is it copywritable? What happens when the client spots that 'copy' in another work, when they think they have an original work with exclusive rights?... That potential come-back bothers me. Not to mention how I will feel when the faces of my family taken from my Facebook galleries start appearing on toilet roll ads... Secondly, I really don't like the idea that Adobe can say 'NO' to whatever I want to create. What the h*** has it got to do with them? 'Computer says no'!... It's bad enough that my lenses are computerised, without my editing software starting to preach morality! Bah humbug! ;)
chat-gpt has that same problem. For some reason they think they can 'limit' the potential for morally / ethically questionable things by preventing 'bad commands', while at the same time not even beginning to answer the big question : whose source material is being used and are they getting fair compensation ? It's tricky enough as is for samples in music/audio, because humans are doing that (for now). I can't even begin to imagine the potential for ethical/moral issues when 'the machine' has done this with zero proof of source material.
according to PHLEARN's recent post about this same subject, Adobe is only using images from its massive collection of Adobe Stock, as opposed to random harvesting from the internet.
@@thatericguy1 Adobe Stock doesn’t hold the copyrights to independent Adobe Stock contributors’ work, so it isn’t really their own library. From what I understand, Adobe Stock contributors have not to date been compensated by Adobe for the use of their photos to train the AI software. As a related aside, an artist’s share of the Adobe Stock licensing fee = 33-35%, and Adobe’s share = 65-67%. So Adobe makes almost twice what the artist does for each stock photo/video download - they can afford to compensate. Adobe claims they are working towards a compensation model, but as they say, that ship has already sailed. Adobe Stock contributors can’t opt out of data set training either, so if they want to sell their photos on Adobe Stock, they are basically agreeing to let Adobe use their images for free to create software they weren’t told about in enough detail to make an educated decision up front - software that will eventually replace them. It’s a passive form of pay-to-play, which is still a form of exploiting artists without paying them, only worse because they are using artists to train their replacements. If we think that this won’t replace stock artists to a huge extent, we aren’t paying attention. In my opinion, the right thing for Adobe to have done would have been to set up the compensation model before they used the contributors’ images, clearly described to artists how they planned to use their work (that’s what everyone else has to do when they license an image), and provide an opt out if an artist didn’t want their images used to train the software.
@@thatericguy1 Interesting. Even if that is true of Adobe, it begs a couple of questions, such as whether contributors were specifically asked if they approved their images for AI harvesting. Elsewhere, Getty have an ongoing legal battle with stability AI, who they allege have harvested 12 million images without permission or payment, with AI companies claiming 'fair use'.
Great video Glyn, as always beautifully explained. I have 'played' with an image I took recently and am really amazed at how Generative Fill works. It does beg the question though about any future images that I am viewing, are they 'real' ?!!
That's awesome...Adobe finally found an answer to fixing the issue in the first 2 photos that we have been struggling with for many years! (either cloning or try stretching the photos 😅)
This is fantastic... we will soon be able to create a photograph of a farmer in Ireland without having to travel to Ireland, or even meeting a farmer. Photography will finally be a hobby anyone can embrace.
Great video. I have been playing with this for a couple days now. It’s mind blowing. I realize this is a beta version, but is there anyway to use this as a plugin with LR Classic????
Does anyone know how you change the theme? Like putting a filter over the whole photo? I want to change my photos in a van gogh style, I know its possible on these AI websites, but somehow in photoshop it doesnt work like that
This is a tremendous video! I am a photo retoucher and the new AI is a thrill to have - soon - haven't checked out the Beta, but will now! Thanks so much for your enthusiasm!
Absolutely mind blown tbh. Sod creativity, if A.I can help me stack money by assisting / speeding up my work flow i'm game. I don't care if A.I is making things too easy or taking away creativity.
What you can do with Generative Fill is mind boggling. I haven't made up my mind as to whether I will try it as yet. Basically because when you start adding things to an original photo it is no longer your creaion per se, it now becomes a composite using AI or images from Adobe stock files.
Great video. Really impressive, and I think ya, this is the first AAA studio that has taken the new AI art software and run with it in a pleasing familiar UI. Looks fun too!
It's only a matter of time for the resolution to increase but even now it alldepends on how big an area you work on ... I defy anyone not to think the hair is really his and the stool was really photographed as it was in the final picture.
Well you can get around the Resolution with free open source alternatives, but that wouldn’t cost a shitton per month to not even own the product in the end right?
Guidelines limitations are ridiculous. It's on our computers and it's a tool we bought. We should be able to do whatever we want. I don't expect a message to tell me to stop drawing when I use my brush to paint a bloody scar
I'm trying Photoshop Beta in particular the Generative Fill function to increase the photos all around, I'm doing various tests, I mainly shoot with Canon Eos R5 with 45 million megapixels, with the original photos the artificial intelligence only works on skies or out of focus parts , with detailed subjects it is clear that the resolution is not at all sufficient and consequently unusable. So I tried to drastically reduce the size of the images like 2000 pixels on the long side, in this case the AI works perfectly with any type of image. So I wanted to know what the exact resolution limit of the Photoshop Beta AI was.
@@glyndewis Hi, what does this mean exactly? That Photoshop Beta SI only works at 100% quality with 1024x1024 photos? Do we know in the future with the final version at what resolution it will work?
@@AlessandroChiariniphotographer It's in beta. Hardly think Adobe are going to make full-resolution capabilities possible in free beta version. Once it's fully developed, I'd imagine a couple of things: firstly, better resolution. Secondly, you pay for it.
I love it for taking lens glare off the subjects glasses, and taking tourists out of travel photos. I also find it takes people's watermarks (if you use them) out very neat and fast. I tried 3 different watermarks. One of them I got the warning but the other 2 worked perfectly. Of course I wouldn't do this to anyone's property for real but I wanted to see if it would work. Also as I mentioned before, it doesn't add clouds to a panorama image too well. There is way more positives about it going forward that the negative. Looking forward to Adobe putting it in PS permanently.
A decade ago, at one of astrophysicist Neil deGrasse Tyson's lectures, when asked about UFOs, he joked that currently Photoshop probably has a built-in "UFO button" ;) Now we really have it.
I'm also amazed by this new Photoshop feature, of course, and I've already used it for (incredibly) fixing some pictures. However, I'm finding an important issue: the low resolution of the AI generated images. At least, for my Nikon Z7II resolution, it becomes unusable many times, it just looks blurry and lacking detail compared to the original file.
this is scary stuff Glyn. I love it as it will make my job as a content creator easier but there is also a scary side to it where thousands of people will be out of jobs.
We've been testing the fill feature and I have to say it's pretty amazing. We actually couldn't believe how accurate the fill is. We had a vertical image of two subjects in a train station (surrounded by trains on either side and wanted to see if it could fill for a horizontal. Well, it filled in the trains and walkway perfectly and even added correct colors/grading, lighting and DOF. We also tested it out with an image of a couple during fireworks on a balcony with hundreds of people below BUT there was an annoying table with drinks and a mans leg in the shot (couldn't ask him to move). The fill perfectly rebuilt the railing with spindles, filled in people below and added directional lighting. Amazing. I'm not a big fan of Adobe and their politics but on a purely technical level this is a huge positive development.
Its about time Photoshop caught up to our creative genius, way to go team!
The stool example was mind-blowing to me! So much fun ahead for us!
Incredible huh 😃
I absolutely love your education style. You always deliver. Thank you
Thanks so much Charles
Absolutely love the Generative fill, used in on many of my photos already!
👍🏻
It's insane what this and AI generated artwork can do. And this is only just the start of it - It's still hit or miss sometimes, but in a year or two it'll be so much better, and it's already pretty amazing.
What I love most about it (other than speeding edits up by a lot, and making previously impossible things now possible), is that it does allow so much more control over creativity. Things you couldn't do before or could only imagine in your head, are now much more achievable. Previously my skills might not have been good enough to get exactly what I wanted, or to fix things that really annoyed me about a photo, but this literally does it in seconds, and usually way better than I could have done it.
Combining this with Midjourney means it's a pretty exciting time for creative work - but it does also mean you can no longer trust any image you see as being real.
Great to hear you find the A.I. helpful
I don't think I'd call it creative work per se.
It's more like uncreative work.
The AI software is doing the "creative" part.
Instant gratification (especially when used by non-creatives), with AI being the workhorse part of the art or photography, will just subdue real human creativity by laziness and lack of pursuing skills and fresh, new ideas and concepts.
Photography is now more like painting than ever before. Creating the changes is not "fake / false" but merely allowing photographers to become creative painters. This is beyond impressive and opens massive improvement potential for any "creator." Thanks for your video.
m.th-cam.com/video/BwmAPSXgXbA/w-d-xo.htmlsub_confirmation=1
@ANASTASI LINE does painting only become painting when you get dirty? What if you use paint while staying clean? And what if your mouse is so dirty that you get messy while using the computer? Hmm. Besides, there are many paintings known by well known painters which are "painted over". So that is no argument.
@ANASTASI LINE DID I EVER SAY ANYTHING THAT CONTRADICTS YOUR STATEMENTS?. AND WHY ARE WE TYPING IN CAPITALS?
eventually these tools will be so good you won't need anyone.
@ANASTASI LINE and what does this have to do with anything beeing said? :D
Integrating this with Photoshop makes the whole graphic and visual design scene even more crazier . Let see where this takes us . Results are really astonishing
Enjoyed this video, thank you! Saw your Photoshop Summit tutorial, looking forward to the next one!
Thanks a lot
Excellent video. Thank you.
Thank you
Amazing. I'm so excited to play with it more - I've already used the ai generative fill to expand photos and it's truly mind-blowing. Would love to see more videos for inspiration.
Thank and yeah no problem ... will be done
I love it. Our design team does as well. We’re always asked for extracurricular things to spice up images in post, or remove very difficult things like cranes, forklifts, woodpiles and tricky license plates. This has greatly sped up mock-ups and workflows. It’s light years better than content aware algorithm and this is still in its infancy, let’s see how accurate it gets 2-5 years from now.
I spent some time learning this new feature in Photoshop yesterday. I love it and am excited to see how it will evolve. This is another excellent tool I will use when needed, and understand it is limited so I will always have to put my finishing touches on the final image. I had some things I wanted to accomplish with Midjourney and could not with its deep learning curve. This PSD AI feature is a walk in the park. Of course, there will be those who will exploit it and think they are now real artists by using it by creating tacky art, just like the millions who believe they are real photographers using their cell phones & DSLRs on auto mode.
FYI, real photographers often use auto exposure modes and autofocus on their DSLRs.
Great experimentation here, Glyn. I love how it handles the shadows and lighting from original image. Also a great way of adjusting to somehing that seems obvious is here tostay. It truly is the creative mind that produces the final image and AI is just another, albeit phenomenal tool.
100% agree with you there
You can tell how advanced of a leap it was because the first reaction from everyone was to get scared... like we'd just seen fire for the first time and it looked like magic.
I used the generative fill on a recent composite of 9 dogs. Extending the background in the same way you did on your farmer portrait.
So brilliant!
Nice one 😃
Thanks Glen, one of the best demos of this AI tool I have seen.
love your charisma and teaching style
Very kind ... thank you
While the assertion you mentioned emphasizes the role of our minds, hands, and tools in expressing creativity, there are indeed other perspectives to consider. Here are a few alternative viewpoints:
Environment and Inspiration: Creativity can also be influenced by external factors such as the environment we inhabit and the experiences we encounter. Our surroundings, interactions with others, and exposure to different cultures and ideas can all contribute to our creative expression.
Emotion and Intuition: Creativity often emerges from a deeper level of emotional and intuitive understanding. It is not solely a product of our cognitive processes or the tools we use. Our feelings, instincts, and personal experiences can play a significant role in shaping our creative output.
Collaboration and Interactions: Creativity can be enhanced through collaboration and interaction with others. Engaging in discussions, brainstorming sessions, and receiving feedback from peers can foster new ideas and perspectives, transforming individual creativity into collective creativity.
Constraints and Limitations: Sometimes, limitations and constraints can actually fuel creativity. When faced with challenges or restricted resources, individuals may find innovative ways to overcome these obstacles and generate unique solutions or artistic expressions.
Imagination and Vision: Creativity often originates from the power of imagination and the ability to envision possibilities beyond the present. It involves the capacity to think beyond what currently exists, to dream, and to visualize novel concepts or artistic creations.
Inner Exploration: Self-reflection, introspection, and personal growth can also contribute to creativity. Exploring our own thoughts, beliefs, and emotions can lead to self-expression and the generation of original ideas.
It's important to note that creativity is a complex and multifaceted concept, and different individuals may have unique perspectives on what fuels their creative process. The interplay between our minds, hands, tools, environment, emotions, and interactions ultimately shapes how we express our creativity to the world.
Generated by ChatGPT - but I pretty much agree :-)
I love it. Gives you complete creative freedom to express yourself in ways that are now less limited. I don't have huge amounts of money for a studio, so using AI can really plug that gap for me.
Great that this technology is going to be so useful for you 👍🏻
Brilliant video! You're spot on with your comment at the end. I've had ideas floating around in my head for years that I just don't have the time or energy to produce due to having to learn new several new software packages extensively. Now I can take my main skills of PS compositing and color grading and combine them with Ai to produce these art pieces. It's about using the tools that we are giving to our advantage.
100% agree with you there. Glad you liked the video 👍🏻
Very cool video! Well done!
Thanks alot
This new fill is amazing! Photos just took a huge step forward.
I agree wholeheartedly with your summation. Thanks Glyn.
Thank you Pip; interesting times ahead
I haven't had a chance to try the Generative Fill yet but Unmesh Dinda (Piximperfect) states that the fill resolution is limited to a cap of 1024 pixels on the long side which implies that you need to keep you "fill" selection areas within this "cap" to maintain the resolution of bigger images. The examples you've shown don't seem to address this Glyn - so I'd be interested in your view on this issue.
In the wider context, whilst I agree that creative images makers will use these tools to add to their creations - my concern is that Ai technology will be another big step in driving photography and design down to the lowest common denominator, with the result that it will become even more difficult for professional photographers and designers to make a living. As a book cover stock photographer it concerns me that the market will eventually be inundated with Ai generated images and designs that have never, as it were, seen the inside of a camera but generated from text prompts alone. There has already been one high-profile example of this where Bloomsbury used an Ai generated image on the cover of "House of Earth and Blood".
When I was a production artist I can think of spendiing two days retouching a photo to clean up a beach to be used for a big boating campaign. With this generate function, I am thinking less than an hour would have been needed. This is powerful for things like that. For that this is beyond good. For the creative mind this is exponential .
But it's also two days of paid work lost.
@@saiyaniam on that but then you can pick up new work
@@mickeydeneher2301 or the Indian kid who can use AI and live cheaply will completely undercut you
@@saiyaniam oh well
Thank you, Glyn for the video and the explanations. 🌷
You're welcome ; thanks for watching
You chose the exact word to describe the stool and floor generation - INSANE!!!!!!!! Many thanks!!!!
😃👍🏻👍🏻
I think it is great. Although it blurs the line of what is real and what is not ... soon enough we won't be able to tell the difference. I wonder what effect that will have in how we perceive our reality and how we process it. Video is next ...
So excited to try out generative fill! I was impressed with kahmas ability to generate stunning AI profile photos.
this is insane, I have been playing with it with mixed results, the good results are really crazy good, the bad results are comically bad, but of course its in Beta, once it has been improved and understood with user submited data it is going to be insane
Wow, you've got some great use cases there. At this point it's just a tool to augment our own creativity. It's great for cleaning up images, and obviously for so much more. I'm looking forward to seeing how it progresses, but in my mind it will never be the end of photography. A large part of photography is just being out, camera in hand, and I can't replicate that with AI.
Exactly that
@@glyndewis Yeah but no one will care if it's taken outside or generated. AI is here to replace the craft, anyone that thinks otherwise is delusional. Just give it time and we'll all be replaced.
Your voice is so well mixed! 😂 Fun video thank you! 🎉
That was Great 🤩👍
Amazing!! This a practical use... I love it!
Absolutely game changing
Pretty impressive, Glyn.
Writing prompts and continously clicking until the rsult fits, is creative?
No you misunderstand ... using it in conjunction with edits 'is' creative ... you saw all the layers too right? 😉
@@glyndewis Clicking and clicking again and again is also not creative in editing. It´s a useful tool for layouting, when the photo is cropped too tight, but I cant see any creativity in turning portrait to landscape.
Love it! Very creative and to think you can repair our treasured photos that were damaged, so great!
Does the AI feature require an online connection?
I agree with the last part for someone who are creative and professional its an added tool, for others will be just filling their portfolio with unrealistic photos, and people by their nature will not care to distinguish between authentic photos that uses the tool to enhance or just completely generated by AI. The value and story behind a photograph will not matter as AI eliminates the hassle of going out and shoot your photos ( in majority of none photographer’s mind set)
Initially, I started to feel threatened by AI fearing that i would be out of work very soon. But then i decided I'm still the one in charge of creativity, and I'm just going to make AI my bitch, and consider it another powerful tool in the box.
@@DavidStarkdyvarg Completely agree with you here.
Amazing what technology is evolving into!
Sure is
Slightly depressing that I use a lot of negative space purposefully in my wedding shots, and now everyone just does it in post 😂
I've been experimenting with it to change or extend backgrounds; I've made a blank wall slightly less blank by adding some faded graffiti to it; and, most egregiously, I livened up a boring corner by materializing a cat in it (Generative Fill did an excellent job on the cat.) But here's the kicker: This technology depends on the machine-learning model being "trained" on huge sets of prototypical images, which it slices and dices to execute your generative prompt. So the more your concept is commonplace, mundane, and obvious, the more likely the MLM is to have been trained on something similar, enabling it to produce excellent results. Meanwhile, the more idiosyncratic, unique, and distinctive your original concept, the less likely the MLM will find similar source materials in its training set, and the more likely it is to produce an "off"-looking result. You can almost use it like a naffness meter -- the more it improves your image, the weaker your original idea was in the first place...
The stool one BLEW MY MIND
Yeah me too 😃
Very Nice 😮
Thanks Ricardo
Hi Glyn yet another great video. Can you tell me if the Remove tool uses pixels from the original image and not add pixels it’s got from else where. I do know you don’t need the internet to use it so it’s not using the cloud to generate the pixels for the image. I’m just researching the ethics of the AI generative tools for competition purposes. Thanks
Need a new computer to access features. I ordered it yesterday and will take 2 weeks to arrive. Hopefully Photoshop does not become obsolete before then, and humans manage survive Ai development
The world is changing at quite rapid speed my friends. This is beyond game changing.
It sure is 😃
@6:53min what type of Camera are you using ???
adding fog etc with A.i sold it for me! I always feel like I need a fog machine or aerisol can. Can't wait to use it! :D
photoshop is magic more than ever!
Sure is
I'm glad you did this video. It's helped me do better photography. I'm now going through all my photos and deleting every one of them where I borrowed unneeded help from other photos. I want to be REAL.
In the portrait of the farmer, the generated areas of rock & grass lack what is important in those areas of the original, the fine gritty texture. Simply adding more stuff that resembles the original portion says that the artist didn't understand what was of value, ie what made the look valuable. In fact were the final version the original, it would have been improved by cropping back to the vertical version. So while this is a great tool it does nothing for visual judgement.
You do understand "beta"?
Ive been using filler textures forever, this changes everything. Sign me up Adobe!
Excellent break down
Yeah, generative fill is a total game-changer. I've worked with retouching for almost two decades, and this honestly blows my mind.
It's a game-changer for those who are playing the game of manipulating their photos into illustrations.
@@yeohi But what about those, who you seem to have ingnored, who "manipulate" their photos ever-so-slightly, well short of 'illustration'? Is it not a game-changer for them too?
WOW!!! that is astonishing almost like wizardry
It's worth noting that Generative fill only does 1024x1024 pixels and stretches the result. So large backgrounds can be grainy and artifacty - although there is a workaround which is easily internet findable. I've found this handy for shots where I haven't quite left enough background and need more space in the photo.
I'm finding AI a bit of a giant everythingburger - if you don't know what you're trying to achieve you'll end up with everything, and kind of, nothing...
Yeah I've done a video about the 1024 x 1024
You've made by far the most compelling video on Adobe AI Generative Fill, taking all the really important perspectives into account. No drama, just advantages with a philosophical ending. Very very nicely done, thanks.
I just used the remove tool to remove a blurry robin from in front of my bird photo of a juvenile starling. using the remove tool. I then used generative fill to generate a new tail for my bird. After a second try, I got a near perfect result!
Superb
as a composite photographer this will make my life easier.
Yes!
Nice one 👍
I've always used alternative software to Photoshop, but honestly this is a game changer and I'm considering switching. Are there any requirements to use this tool? Also, do you need to pay extra to enable it, or is it enabled with an ordinary Photoshop subscription?... And, is there a feature to add objects to a photo? I've recently started making framed wall art mockups using stock photos of interior rooms, but it's so time consuming to make the frame blend in with the room and look realistic. If I could just write a prompt to add a blank frame on a wall, and if it looks as realistic as the chair's legs, that would save me like 99% of the time I spend doing it manualy
Yes you could select a specific area and then prompt for a blank frame in the space.
This sounds too good to be true, cant wait to try it out myself and see the magic.
I love the creative opportunities and the ability to enhance one's own original work.
The main thing that concerns me though (with AI generation generally, not just Adobe's Generative Fill) is where the replacement / additional elements are being harvested from other photographers' work. For instance, were all the hair options for the viking created from scratch by Photoshop? Far more likely they were derived at least in part from the work of another unknown, unpaid photographer. Next time it may be your work making-up someone else's photo. Is that saleable as an original work? Is it copywritable? What happens when the client spots that 'copy' in another work, when they think they have an original work with exclusive rights?... That potential come-back bothers me. Not to mention how I will feel when the faces of my family taken from my Facebook galleries start appearing on toilet roll ads...
Secondly, I really don't like the idea that Adobe can say 'NO' to whatever I want to create. What the h*** has it got to do with them? 'Computer says no'!...
It's bad enough that my lenses are computerised, without my editing software starting to preach morality! Bah humbug! ;)
chat-gpt has that same problem.
For some reason they think they can 'limit' the potential for morally / ethically questionable things by preventing 'bad commands', while at the same time not even beginning to answer the big question : whose source material is being used and are they getting fair compensation ?
It's tricky enough as is for samples in music/audio, because humans are doing that (for now). I can't even begin to imagine the potential for ethical/moral issues when 'the machine' has done this with zero proof of source material.
according to PHLEARN's recent post about this same subject, Adobe is only using images from its massive collection of Adobe Stock, as opposed to random harvesting from the internet.
@@thatericguy1 Adobe Stock doesn’t hold the copyrights to independent Adobe Stock contributors’ work, so it isn’t really their own library. From what I understand, Adobe Stock contributors have not to date been compensated by Adobe for the use of their photos to train the AI software. As a related aside, an artist’s share of the Adobe Stock licensing fee = 33-35%, and Adobe’s share = 65-67%. So Adobe makes almost twice what the artist does for each stock photo/video download - they can afford to compensate. Adobe claims they are working towards a compensation model, but as they say, that ship has already sailed. Adobe Stock contributors can’t opt out of data set training either, so if they want to sell their photos on Adobe Stock, they are basically agreeing to let Adobe use their images for free to create software they weren’t told about in enough detail to make an educated decision up front - software that will eventually replace them. It’s a passive form of pay-to-play, which is still a form of exploiting artists without paying them, only worse because they are using artists to train their replacements. If we think that this won’t replace stock artists to a huge extent, we aren’t paying attention. In my opinion, the right thing for Adobe to have done would have been to set up the compensation model before they used the contributors’ images, clearly described to artists how they planned to use their work (that’s what everyone else has to do when they license an image), and provide an opt out if an artist didn’t want their images used to train the software.
@@L8chasr very good points. Here’s hoping the compensation model evolves to a more satisfactory solution for Stock contributors.
@@thatericguy1 Interesting. Even if that is true of Adobe, it begs a couple of questions, such as whether contributors were specifically asked if they approved their images for AI harvesting. Elsewhere, Getty have an ongoing legal battle with stability AI, who they allege have harvested 12 million images without permission or payment, with AI companies claiming 'fair use'.
Brilliant, What about the "not for commercial use" agreement?
Great video Glyn, as always beautifully explained. I have 'played' with an image I took recently and am really amazed at how Generative Fill works. It does beg the question though about any future images that I am viewing, are they 'real' ?!!
That's awesome...Adobe finally found an answer to fixing the issue in the first 2 photos that we have been struggling with for many years! (either cloning or try stretching the photos 😅)
Great video. Thanks.
Interesting to think about the possibilities and scary at the same time
This is fantastic... we will soon be able to create a photograph of a farmer in Ireland without having to travel to Ireland, or even meeting a farmer. Photography will finally be a hobby anyone can embrace.
Thankfully this farmer was in wales so no worries there then 😉
@@glyndewis command+Z
Great video. I have been playing with this for a couple days now. It’s mind blowing. I realize this is a beta version, but is there anyway to use this as a plugin with LR Classic????
Does anyone know how you change the theme? Like putting a filter over the whole photo? I want to change my photos in a van gogh style, I know its possible on these AI websites, but somehow in photoshop it doesnt work like that
great video i sub
Thank you
Is this in the neural filters option or in the normal filter menu? Thanks, Looks very cool
It's currently only in the public beta version of Photoshop, and it appears in a "context bar" which also is new in the beta version.
what version of photoshop have this tools?
This new feature blew me away. But you aint seen nothing until you try kahma!
This is a tremendous video! I am a photo retoucher and the new AI is a thrill to have - soon - haven't checked out the Beta, but will now! Thanks so much for your enthusiasm!
Absolutely mind blown tbh. Sod creativity, if A.I can help me stack money by assisting / speeding up my work flow i'm game. I don't care if A.I is making things too easy or taking away creativity.
Insane!
I love how we call "our creativity", and "our minds at work, using a tool" we simply picking some suggestions a software gave us
Only if the entire image is A.I.
What you can do with Generative Fill is mind boggling. I haven't made up my mind as to whether I will try it as yet. Basically because when you start adding things to an original photo it is no longer your creaion per se, it now becomes a composite using AI or images from Adobe stock files.
Great video. Really impressive, and I think ya, this is the first AAA studio that has taken the new AI art software and run with it in a pleasing familiar UI. Looks fun too!
Glyn do you think the restrictions in resolution for what AI produces, is just the beta version, and will not be limited in the full version ?
It's only a matter of time for the resolution to increase but even now it alldepends on how big an area you work on ... I defy anyone not to think the hair is really his and the stool was really photographed as it was in the final picture.
@@glyndewis Thanks Glyn - Interesting times we live in !!
Well you can get around the Resolution with free open source alternatives, but that wouldn’t cost a shitton per month to not even own the product in the end right?
Is it better to work on a raw file coming from camera raw, a tif file 8 bits or a jpg?
Whatever file you want; carry on editing as normal and use this 'as and when' ... it doesn't replace all the editng before it.
Guidelines limitations are ridiculous. It's on our computers and it's a tool we bought. We should be able to do whatever we want. I don't expect a message to tell me to stop drawing when I use my brush to paint a bloody scar
Beta
I'm trying Photoshop Beta in particular the Generative Fill function to increase the photos all around, I'm doing various tests, I mainly shoot with Canon Eos R5 with 45 million megapixels, with the original photos the artificial intelligence only works on skies or out of focus parts , with detailed subjects it is clear that the resolution is not at all sufficient and consequently unusable. So I tried to drastically reduce the size of the images like 2000 pixels on the long side, in this case the AI works perfectly with any type of image.
So I wanted to know what the exact resolution limit of the Photoshop Beta AI was.
1024 x 1024 in Beta
@@glyndewis Hi, what does this mean exactly? That Photoshop Beta SI only works at 100% quality with 1024x1024 photos? Do we know in the future with the final version at what resolution it will work?
@@AlessandroChiariniphotographer It's in beta. Hardly think Adobe are going to make full-resolution capabilities possible in free beta version. Once it's fully developed, I'd imagine a couple of things: firstly, better resolution. Secondly, you pay for it.
Great!
Possible on ipad pro?
I love it for taking lens glare off the subjects glasses, and taking tourists out of travel photos. I also find it takes people's watermarks (if you use them) out very neat and fast. I tried 3 different watermarks. One of them I got the warning but the other 2 worked perfectly. Of course I wouldn't do this to anyone's property for real but I wanted to see if it would work. Also as I mentioned before, it doesn't add clouds to a panorama image too well. There is way more positives about it going forward that the negative. Looking forward to Adobe putting it in PS permanently.
A decade ago, at one of astrophysicist Neil deGrasse Tyson's lectures, when asked about UFOs, he joked that currently Photoshop probably has a built-in "UFO button" ;) Now we really have it.
Tried generative fill, so cool! I liked kahma io results for LinkedIn pictures just as much.
Hi Glyn, was curious if you've compare Topaz Gigapixel vs Adobe Photoshop's Upscaling.
I was on the right track! Lol
I'm also amazed by this new Photoshop feature, of course, and I've already used it for (incredibly) fixing some pictures. However, I'm finding an important issue: the low resolution of the AI generated images. At least, for my Nikon Z7II resolution, it becomes unusable many times, it just looks blurry and lacking detail compared to the original file.
Resolution will increase for sure once out of beta 👍🏻
@@glyndewis Let's hope, it doesn't make much sense right now but for low (or perhaps medium) resolution work.
this is scary stuff Glyn. I love it as it will make my job as a content creator easier but there is also a scary side to it where thousands of people will be out of jobs.
So excited to try out generative fill! I was impressed with kahmas ability to generate stunning AI profile photos.
what about importing video and working on content, i'm downloading it right now but curious to know about the video side of it.
I wonder if this is eventually going to be an extra subscription. I hope it will be included in the photography plan.
#Whoa thats insanely cool. You have to try kahma to make portraits of passed loved ones with AI too.