ADOBE FIREFLY AI: The TERRIFYING New Reality for Artists!
Vložit
- čas přidán 28. 08. 2024
- Adobe Firefly AI, the new ai tool suite from Adobe was recently released and this lanches a brand new reality for artists that is gonna change the way people create! In today's video, I'm gonna show you Firefly's features, what we can expect from that tool and how their ethical image sourcing and training will give a reality check to everyone that is against ai art! So let's go!
What do you think of Adobe Firefly? Let me know in the comments!
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
SOCIAL MEDIA LINKS!
✨ Support my work on Patreon: / aitrepreneur
⚔️ Join the Discord server: bit.ly/aitdiscord
🧠 My Second Channel THE MAKER LAIR: bit.ly/themake...
📧 Business Contact: theaitrepreneur@gmail.com
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Runpod: bit.ly/runpodAi
Adobe firefly: www.adobe.com/...
firefly.adobe....
All ControlNet Videos: • ControlNet
My previous ControlNet video: • TRANSFER STYLE FROM An...
3D POSE & HANDS INSIDE Stable Diffusion! Posex & Depth Map: • 3D POSE & HANDS INSIDE...
Multiple Characters With LATENT COUPLE: • MULTIPLE CHARACTERS In...
GET PERFECT HANDS With MULTI-CONTROLNET & 3D BLENDER: • GET PERFECT HANDS With...
NEXT-GEN MULTI-CONTROLNET INPAINTING: • NEXT-GEN MULTI-CONTROL...
CHARACTER TURNAROUND In Stable Diffusion: • CHARACTER TURNAROUND I...
EASY POSING FOR CONTROLNET : • EASY POSING FOR CONTRO...
3D Posing With ControlNet: • 3D POSING For PERFECT ...
My first ControlNet video: • NEXT-GEN NEW IMG2IMG I...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
►► My PC & Favorite Gear:
i9-12900K: amzn.to/3L03tLG
RTX 3090 Gigabyte Vision OC : amzn.to/40ANaue
SAMSUNG 980 PRO SSD 2TB PCIe NVMe: amzn.to/3oBR0WO
Kingston FURY Beast 64GB 3200MHz DDR4 : amzn.to/3osdZ6z
iCUE 4000X - White: amzn.to/40y9BAk
ASRock Z690 DDR4 : amzn.to/3Amcxph
Corsair RM850 - White : amzn.to/3NbXlm2
Corsair iCUE SP120 : amzn.to/43WR9nW
Noctua NH-D15 chromax.Black : amzn.to/3H7qQSa
EDUP PCIe WiFi 6E Card Bluetooth : amzn.to/40t5Lsk
Recording Gear:
Rode PodMic : amzn.to/43ZvYlm
Rode AI-1 USB Audio Interface : amzn.to/3N6ybFk
Rode WS2 Microphone Pop Filter : amzn.to/3oIo9Qw
Elgato Wave Mic Arm : amzn.to/3LosH7D
Stagg XLR Cable - Black - 6M : amzn.to/3L5Fuue
FetHead Microphone Preamp : amzn.to/41TWQ4o
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Special thanks to Royal Emperor:
- Totoro
Thank you so much for your support on Patreon! You are truly a glory to behold! Your generosity is immense, and it means the world to me. Thank you for helping me keep the lights on and the content flowing. Thank you very much!
#stablediffusion #adobe #firefly #aiart
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
WATCH MY MOST POPULAR VIDEOS:
RECOMMENDED WATCHING - My "Stable Diffusion" Playlist:
►► bit.ly/stabled...
RECOMMENDED WATCHING - My "Tutorial" Playlist:
►► bit.ly/TuTPlay...
Disclosure: Bear in mind that some of the links in this post are affiliate links and if you go through them to make a purchase I will earn a commission. Keep in mind that I link these companies and their products because of their quality and not because of the commission I receive from your purchases. The decision is yours, and whether or not you decide to buy something is completely up to you.
HELLO HUMANS! Thank you for watching & do NOT forget to LIKE and SUBSCRIBE For More Ai Updates. Thx
Bor i love you 0:07
To me personally the safety of it being a model that claims to have rights to all images used to train it and the ease of use and no longer haveing to fight with python and scripting shells I might be sold for 20/month
As someone who studied ethics in two fields (phil, and psych), and then dabbled in auxiliary fields (game theory, econ, bio-psych, evo-psych). Often, the perceived difficulty is just pre-scientific ideals of morals. 95% of it vanishes when taking into account goals-directed thinking being mostly stimulated below the conscious level. The other 5% is genetic variance, which by this point is trackable. Preliminary research suggests that genetics mostly breaks into 3 sets of thinking A) balanced ethical values, B) overvaluing care C) overly literal thinkers. The issue is that AI art threatens both A & B, leaving the extreme minority of C being the supporters.
-> They didn't back the open source free for all version and now they must fight the corporate legally protected version. I love the irony. Reject teamwork, join true evil. I talked to a few people and their response was "Fuck adobe. Pirate them and use their tech." so the artists found an excuse to join the AI art now. They have an excuse now. In fact, since they stole their version they'll keep shittalking us for being less moral than them.
@blaze armoru honestly your right the corporate model looks like it's gonna be a winner and we will unfortunately be greatful for the crumbs they allow us now. That said the Adobe version hits the ease of use barrier and the usefulness thresholds for it being viable so it'll succeed and between Adobe google and Microsoft they'll be the ones in control of all the software that gets to consumers. It's all heading that way already.
I am actually waiting what Adobe's response is to AI art. And well they created their own streamlined, powerful, easy to use, fully integrated and "ethical" Automatic1111. Fantastic.👍
I was actually impressed for the same reason you were unimpressed, haha
I didn’t have high expectations for Adobe to implement these tools this effectively. I expected a simple text-to-image generator inside of Photoshop. Maybe doing some in-painting eventually…
The fact that all the greatest tools we’ve been using for months have suddenly had the barrier-for-entry removed and placed in front of virtually every professional creative individual is amazing to me.
This is a rare A+ for Adobe imo
I thought they had given up lol. 😂
agree. this feels like a massive leap in UI/UX, instead of having to watch youtube tuts to instal SD locally so i can install a bunch of stuff to type prompts into a powershell window then check an output folder. this is automatically lightyears ahead. i love tech as much as anyone, but UI matters, and this is a solid interface i just hope it has a dark mode.
@@itsjustcavan Exactly, people underestimate how important UI/UX is. This feels like all the tools we've been playing around with but simplified so even my aunt could create cute cat pics and send it as postcard. Many research always claims less is more when comes to UI. StableDiffusion has always been more for the explorative crowd trying to build workflow stacks for individual use or scaled up apps such as Midjourney or Lexica that have StableDiffusion under the hood basically.
Yeah it’s actually pretty bad ass. Can’t wait for the free rip off of it 🥹
@@justanotherhuman3668 That is just Stable Diffusion, friend 😂
Artists who incorporate AI into their workflow will replace artists who reject AI.
Anyone can be a artist now
exactly
depends, I would love a good AI for consistent photo editing, for example. i.e. skin and hair. Would save ton of time.
For now, regular tools are still better than AI.
In cooperate fields, I think this is 100% true. Everyone else will be regulated to the Indie Market. I think this might have the same effects, CGI did to the film industry when it replaced practical effects. All movie lovers still want and love Practical Effects with traditional models/etc. But that's not stopping big studios from using CGI.
When everyone IS, noone IS....got it?
They missed a huge meme opportunity to name it FireflAi
Damnn 😩😩
As a pro artists (who is pro AI) what worries me here is that this'll lower the bar in terms of quality.
Because these results ARE NOT BAD, but at the professional level they still need polish, the details are wonky, certain aspects are off.
Now if you know a little photoshop it's an easy fix, but I can tell you from experience that the people who make the calls often have no eye for aesthetics and go the route of 'good enough' because to the untrained eye these results are good enough. For example that vector-looking lion at 11:23 has a lot of potential, but if you actually look at it, it's kind of a mess. And while its problems are simple to address, the person in charge may tell you not to bother. So the quality of the graphics you will see in media could potentially go down.
Midjourney already did that: Adobe stock is flooded with really low quality Midjourney images.
Pro artist (and enthusiast) here as well (I think there’s a lot more of us interested in that technology that this channel realizes). I think we just need for the technology and craze to mature.
The overall low bar is a normal step for every new media revolution, ( think early colored films, sound films, early VFX & video games, 3D movies etc…) at first every one gets drawn to the new tool and it gets absolutely swamped with an overwhelming amount of extremely low quality contentent. Uneducated content creators cranking the sliders to the max…
So sure, for now (MJ, SD etc are drowning in meaningless images right now - despite the craze).. But in time as the technology matures and the dust settles we’ll be starting to see more curated content rising again by the people using those tools with more purpose. We’ll probably see some paradigm shift of what constitutes “good art” well and within art history itself. All fascinating.
Lmao imagine complaining about the potential advertisements that I ignore looking slightly off maybe. I guess we should shut it all down.
Hourly rates will drastically go down… if everyone’s an artist now, corporate profits will increase guaranteed
@@snoweh1 If that's all you can imagine this will potentially effect, then your imagination is lacking.
Not fan of Adobe, and technologically not _that_ impressed, but honestly having complete easy to use fast AND legally clean product is breakthrough in itself in this field, one that should not be underestimated. This is a big day for business based in AI imagery.
What Adobe basically did was take all the image generation projects and put them together in one nice interface, improving some of the features they already offered.
And a lot of the AI haters will embrace it simply due to it being Adobe lol
@@Gh0sty.14 not really. Artists famously hate adobe and their software, they just use em cause they have to
@@GerasSB Yeah, but they'll probably end using Adobe AI cause they'll have to, too.
the world of warcraft of AI tools.
been waiting for someone to do this im just sad that its adobe lmao
obligatory "fuck adobe" comment
Lol
Well, they DO have FAR too much power and influence in the digital art industry. There is plenty of reason to hate Adobe. They are too big, they need to be put in their place.
generative AI basically kills stock images anyway.
The usage, Yes. The contributor, No.
Think about if you just upload a bunch of photos with enough quality. No concern much about how interesting or use case of it. Then you get paid by the share of subscription payment people paid for it.
Dataset contributor is a new easier job.
@@ilakya nobody wanna pay dataset contributor though, no sales though the stock - no motivation to upload anything there. You upload hoping for money some "Classic style background", Adobe uses your background to generate AI image for "a rubber duck on a classic style background". Adobe gets money, you receive nothing.
I'm curious what other wierd argument against AI art the haters will come up with this time🤣
"NOOOO! THIS IS TAKEN WITHOUT THE CONSENT! PEOPLE UPLOAD THEIR IMAGE TO BE USE DIRECTLY NOT TO BE TRAINED ON"
They said.
Spoken by someone who’s not an artist and makes a living from it I presume?
@@TheLunarnotesoh, u means AI prompter?
Sidenote: Adobe stock has millions of images created with Midjourney and Stable Diffusion and those are used in the training too. I wonder how the copyright goes with that.
well since as they ruled AI generated art is in no way copyrightable, everything they used from Midjourney is completely legal.
Way to shoot yourself on the foot if you ask me
I'm pretty sure they can filter them with one click of button.
Because the AI doesn’t simply copy/paste from its dataset, the things produced with AI are considered copyright free until transformed by a human
Doesn't matter, AI generated images are not copyright protected.
So? Mid journey and any other txt2img model has that as well, why is it a problem if Adobe does it? Training AI on work without consent of the creator is a fundamental problem in AI anyway.
Exactly what I predicted, now the "artists" that were shitting on open source projects like unstable diffusion will have the pleasure of paying 300 bucks or a subscription to stay competitive in their field instead of using free "available to all" solutions.
They have been brainwashed to become the perfect consoomers ... they think only products by greedy corporations are the only ones that have legitimacy, it's really sad and pathetic.
Why are you using quotation marks around the word? There's already been free alternatives to Adobe for a long time.
the reason we "shat" on open source project is their first database, which contained an unknown numbers of images where their authors didnt gave any consent to be scraped like that.this is for stablle diffusion. for midjourney its even darker, since they didnt released any info on their data taht was used to train, in an era of open source and transparency, it feels really shady. this fact and the ability of ai user to "mimic" or "copy" the style of other...something that when regular artist saw another one copying like that, he's getting labelled publicly, copying a style is viewed as a scumbag behavior if its not for a clear hommage or thanks to the artist who's been copied.
imagine what we see often in prompts... "trending on artstation" "art by X".
we have everything in our hands to blame (not on a court point of view but on moral point of view) generative ai companies, open source or not, or their users.
Most professional artists were already paying for the Adobe suite. This just justifies the subscription cost. Also, a lot of us get it through our employers.
@@markborok4481 also very true!
Aitrepreneur: hates corporations stealing
Also aitrepreneur: it’s okay if it’s decentralised people stealing
Great morals.
Pretty cool how it output as a PSD file with layers for individual objects in the images
The version of Firefly available to the public in the beta test doesn't actually do that, all you get is a flat jpg.
The reason why the copyright of training material is a big deal in my opinion is not because of guilt or something like that.
The fact is, even if you think it's completely fine to use the images this way, the copyright holders still might come knocking on your door, and you don't even know if the court will side with you.
It's not a big deal for an enthusiast, but it's a big deal for a professional.
Two things: Firefly being ethically trained is going to normalize the use of art AI, but may push more laws into place to recognize the difference between "legal" and "infringement" AI art. Even if the ethical argument against AI art is tamed by this option, those that hate AI will still push the narrative that artists will lose their jobs over this... and they are partially right, if you aren't a famous artist and you won't use AI, then your job options will definitely decrease going forward. But we've seen this before, the same stuff happened when cameras were invented, and when digital art tools were released. The landscape changed, but artists didn't go away. I firmly believe that if you have an art department in your company, the people you want running the AI art are artists, since they have an understanding of color, space, and composition that non artists don't have, plus an artist can manually fix flaws in the generated art, which a pure prompt slinger can't easily do since they have to rely on the results of inpainting and controlnet.
I think the real potential here is continued innovation at a breakneck speed. They’ve got the funding and talent to continue improving their tools. I think in two years this suite will be difficult to imagine.
I think it's great that we get an option to do the same thing as we can do with Stable Diffusion, and we also get to pay Adobe for it! /s
What the big media companies and big tech companies are trying to do by opposing Stable Diffusion isn't to help artists, it's to shut out any open source competition, because the big companies already have legal access to all the art they need (and additionally, they don't have to use the legally established fair use right to train AI on public data, because users have explicitly, if unknowingly, given them the license to use it when they signed up for their cloud services).
My thoughts exactly.
Well, let's hope this at least helps open some eyes to the benefits of AI for artists.
as a heavy photoshop user, i really can't wait and excited for this
While the results are nothing new there is a lot of value in having a well designed tool. I mean thats kind of Adobes entire thing. Photoshop was not alone all this time. But gods was it easier to use for a long time then its competitors.
Artists will find something else to hate Firefly for, the issue was never how the images were obtained, it was just their ammunition. The issue was the fact that artists are absolutely scared out of their minds for their profession
I'm not just scared, I literally wish I was never born to begin with, ar died, or had no artistic spark to begin with. I guess, after we all die jobless from starvation, it will be very beneficial for humanity, because it will diminish the amount of mentally ill people, as mental ilness correllates with artistic abilities. Great win for humanity, you can celebrate!
Adobe just legitimized AI. And that's good.
I think AI art has its place, but it won’t replace artists. AI is trained. It doesn’t have a perspective, or experience to put to paper.
I can't imagine artists abandoning Photoshop, because it has AI generation built-in.
I CAN imagine artists scrambling to learn this feature, when they see their collegues productivity going through the roof.
Checkmate, Adobe.
As a guy making his living with art u have to say .... I love it going big into ai ❤
In terms of Text Effects, you can use Depth (either through a Depth model + Depth IO or ControlNet) to get similar effects; Put white text on a black background and feed that in with a prompt and you get similar effects. Theirs might be a little more flexible but should be doing about the same under the hood.
Looking forward for firefly to be honest.
Been using SD for a while and I'm sure that midjourney v5 is still way more powerful than firefly for now but the way Adobe is trying to streamline it into our work flow as artist is maybe not groundbreaking but it's so convenient. Ai is supposed to improve our productivity, not hinder it with slow process.
So I think it will take a while before it's actually usable and useful (like every Adobe product to be honest..) and I sure hope they won't just use that as a gimmick to pump our subscription price higher (who am I kidding..) but if they can give us that tool integrates into light room and photoshop (and premiere, AE etc..) it might actually be game changer for a lot of us.
And I do not agree with your statement about images being free to use because they're online. So the way they're doing it, not ripping off artists from their right is the ethical thing to do.
I'm a stock seller as well so I don't mind my photos being used to train models but if you're gonna use them I deserve compensation. I'm curious as how Adobe compensate the artists though as they haven't for instance bought the use of all my images available on Adobe stock, so I'd be curious to know how they proceeded.
Afterall as you can see the music industry is not letting anyone simply use their songs for free to train their models so why other artists should?
Also remember that for compensation we're probably talking like 30cents per image, artists aren't gonna make any kind of living off those, but it's the least those AI companies can do
Totally agree. Images posted online are free for everyone to see. Acting as if Ai is just coping and pasting is just ignorance.
just cuz an artist posts their work online for you to look at doesn't mean they gave up the rights for you to do whatever you want with it. it's not a free for all. that's why copyright laws exist.
@@pixelpuppy I said it's free to see...not to do whatever you want with it. People see stuff everyday that someone can claim is their copyrighted material does it mean their eyes should be gauged out because they didn't pay for it? People create art and media from inspiration of what they saw throughout their lives. How many artists have created things based on other people's works? My gripe is that these artists think the AI is purely coping and pasting their work when that is simply not the case.
@@pixelpuppy The problem is this: Would an artist that tries to mimic another artists style be at fault? The answer is no, nor are they prohibited from learning by watching copyrighted material.
So then the question becomes: At what point is an AI "an artist" and not "a tool?"
It also opens up a loop-hole that you better bet will be used extensively. You can simply pay an artist to mimic a style you want your AI to learn, and then use those images. No need to touch the original artist if they refuse to allow their art to be learned from.
It's a bit like how you cannot copyright a program, a look, a music instrument, and so on. Trademarks can block the name recognition from being used. Copyright can block straight up copies. Patents can be used to block a new invention from being stolen. (Software patents are a blight, they don't help innovation like patents are meant to and only benefit patent trolls, the top 3 software companies, and the patent lawyers, everyone else lose out on those.)
Just to be clear, AI doesn't copy, it learns. You can prove that by comparing input data to the size of the model. If it was a compression algorithm it would be revolutionary, but it's not. No more than a human brain copies things when it learns, anyways.
@@Sylfa Legal Eagle actually has a video talking about this exactly. And the answer isn't as clear cut as the analogies you're giving. There are a lot of complex differences that make a lot of this still up in the air or debatable. I get what you're saying, but it's not that simple.
firefly = stable diffusion
I think could be stronger
@@NERV0USMONKEYS Adobe will charge for its interface for AI noobs
@@orion4d727 I don't know of any Ai who plan to create layered vector files
@@NERV0USMONKEYS For the moment this tool is only an advertising video, otherwise there are several projects in development (vectorfusion for exemple)
Nice. The fight for copyright has always seemed strange to me, especially in the context of talking about artists who draw existing characters or people without asking permission, but are shaking over each time someone use they art even without commercial intent.
I'm an artist myself but I I really think that the fight for copyrights harms the art community itself, reduces the number of mentions of the unpopular artists, and prevents the art from spreading.
In this situation, only super popular artists win, but if your drawing is on your blog, no one knows about it and you don't want for people to take it and use it in different fan groups, you want all your likes to remain only on your blog... Congratulations you won't get likes at all.
Now, with the advent of artificial intelligence, things have become even more complicated. :D I am very happy with the fact that people got more opportunities to illustrate their fantasies. As an artist, I'm not afraid of losing my job, because artificial intelligence is obviously just a tool that can help get better results. And about copyright, this is ridiculous. Just because ai isn't copy any art but simply learning how to draw analysing it. Well.
It is obvious that the washerwomans are trying to fight against the washing machine. We already know the result because who now knows at least one washerwoman?
P.S: oh and yes, about this Adobe ai... Nice paid thing for people who like to pay for things and who don't like to share something with others for free. Lol.
Every photoshop user before : that’s bullshit this will destroy everything.
Every photoshop user now: that’s impressive I will use it😂
Celsys, makers of Clip Studio Paint, tried to incorporate Stable Diffusion into their software. After a lot of outcry from their user base, they removed it. Although they also didn't have an "ethically-sourced" model they were using, so Adobe's case is different.
I’m interested to see what argument the Luddite crowd jumps to now that their “it’s theft!” argument has been so thoroughly eviscerated.
That's how you can tell it was a psy-op to give government more control over this new technology -- the argument was based entirely on lies.
I would love to see an open source version with an interface like this. Stable diffusion has great features, and all but the interface is really not user friendly, and if your laptop doesn't have the hardware, image generation and workflow is so slow that it gets frustrating. What I love about this release is that big corporations and their monetization of free tech push the open source community to develop better UI and UX. I hope a group of developers get infuriated enough by Firefly that they use this as inspiration to develop an exact replica and call it something more catchy.
oh sweet irony... but honestly, this tech looks really impressive! it seems to have a bit of controlnet, a bit of the slidersfrom artbreeder. very intersting for sure. lets hope we can import custom models to it, so we can make images on all artstyles we want, and not just glorified stock photos.
Didn't Adobe just last year automatically set the permission for user's content to be used for AI training or do I remember incorrectly? "Adobe May Be Using Your Photos to Train its AI". So maybe they weren't so ethical, or were they, if it was opt-out, not opt-in.
User's content training was for AI tooling inside Photoshop, like smart brushes. So as far as I'm aware, it wasn't for this specific AI, but for locally used tools that use some user-related training. Just was blown out of proportion.
Yes that is true, so much for consent. But you should also always read the small prints I guess?
I don't think anyone would claim Adobe to be an ethical safe-haven. But it's legal. And open-source options can follow the same path, more or less.
Instead of learning the style of Pierre Bonnard you could for instance learn the style of Claude Monet, and then use that style to draw "intimate domestic scenes." It would be harder, but as each artist learned from the masters before them you could absolutely re-create the various modern styles by going back to the masters they drew their inspiration from.
Also… It's legal to pay an artist to draw something "in the style of ___" and have them release the copyright to the one paying for it. Giving you perfectly useable art to train an AI on. In a way you're basically just paying a domain expert to create training data, it's the style and to get enough examples of useable subjects/adjectives/etc that is needed. It doesn't *have* to be able to make a decent Monet replica to be useful.
It's interesting that all the pictures in the presentation don't see fingers...
You are roasting nobody; those are not non-sensical arguments. You just don't want to accept the unethical reality of the scraping models. What Adobe did was morally and legally the right thing to do, and what should have been done from the very beginning by everyone else, even the open-source models. Honestly, the only reason I don't use Stable Diffusion is because it's not ethically right. You having a different/non-existent moral compass doesn't make it non-sensical.
Let's be honest here. It's not the fact that images are being used as inspiration to create new works - it's the idea that it can be done so easily. Every artist has used previous works for inspiration for all media and every artist has their style created from something previous or current. If you take an AI generated image that was off the backs of previous artists BUT it was created by hand, there's no issue. But the fact it's done so easily is where the fight lies. Artists that have worked YEARS getting to where they are don't like the fact that Joe from wherever that's never drawn anything can now create something as nice as they can within minutes. THATS the fight. I started years ago using artists I admire to recreate what they do and hone in on techniques. And then I would do this with several artists to get something new and fresh for myself. It's the long tiring workflow to achieve something special. SO, in saying that, I personally use AI to create moodboards, briefs and inspiration for a piece that I create separately. But I don't hate on someone using it however they want. There's still the art of the eye and taste behind art. It's a HUGE time savor for myself and I get results sometimes I wasn't thinking about. AI can be a Godsend to artists from every style. Anyway, my rant and my opinion.
I’m not against ai but for you, a non artist and someone who hasn’t ever appreciated the value of art before ai started making pretty pictures you could control, presuming to know whether or not non ai assisted art will die or not is hilarious at best.
A ton of artist love these tools. I am an artist and I love it! My dad is an oil painter and he thinks its awesome!
Many people (sadly, bad schools I guess) don't understand how economics really work. In reality, "increased efficiency" just means less demand for labor (meaning lots of people laid off, remaining salaries lower because too much supply of people competing for few jobs) and more profits for corporations. If "anyone can be an artist" that means that the rarity of being an artist is reduced to zero, and therefore the value is also near $0. "Democratization' just means "demonetization". That's just how capitalism works, whether you agree with it or not.
I kinda hate how Ai people keep promoting this idea of how Ai will save people’s jobs. Your just training a machine to do your job better than you can. Basically training your replacement to take over for you. Channels like this are kinda pointless.
Love the intro xD. Someone should clip it to send ppl your way. Great channel. Merci! (french from where?)
As an artist myself, I am not completely against to the use of AI, and I genuinely appreciate Adobe for their legal use of images. However, we still cannot fully protect our art from illegal usage, which many companies engage in.
And it breaks my heart to see AI being trained to replicate the styles of specific artists. We need common ethical standards that people know to prevent these problems. Right now, people are just crazy about this!
i'm in the same boat as you. been learning how to do AI art, and it's a love-hate relationship. I can see it being a super useful tool, but the ethics and legal ramifications is actually VERY complex. It's not a simple black and white answer right now.
How do you feel about human artists replicating the style of specific artists?
@@xyzzyxyzzy2 I think this is the key question. It's not so much about the tools used , it's about what people do with those tools.
How do you feel about Sam Does Art? He makes profit by blatantly copying styles from others.
Just to be clear, it's perfectly legal for an artist to learn from copyrighted works. As long as they don't claim to *be* the other artist, nor tries to copy their art.
In music there's even a right to perform where any artist can make a version of any other artists song. You have to pay the original artist for the royalty, but they can't block it. And music is some of the most locked down art you can find thanks to record companies trying to stay on top of their money piles. (The artists usually get screwed over quite hard, it's the record labels that gain the most from the laws.)
Firefly is much easier and coherent than stable diffusion.
That's the Achilles tendon from stable diffusion. I have to check 10 videos ( yours usually to know what parameters are good for a model or tool) so i can see what something does. As everything is Lost in a reddit or random web.
This makes AI logical. And for someone like me who has used Photoshop for years is amazing to use AI as Photoshop tools. As Photoshop interface is ages above any modern AI GUI
Stable diffusion needs to work on this
you are muuuuch more limited with firefly though (as of right now)... I wouldn't want to miss all my custom models, loras and some of the control net options we currently have inside of stable diffusion..
This doesn't make AI logical, it makes it braindead simple. What you call SD's Achilles' heel is really just the need for a bare minimum of domain knowledge in order to use some of the most cutting-edge technology available.
Stable Diffusion is not really meant to triumph Photoshop. It's a monolithic tech stack that allows to be extended and customised entirely to your liking. Automatic1111 was one of the early ones that build a GUI on top of StableDiffusion which is know widely known, but don't forget about Invoke AI which is also open-source. Midjourney implemented StableDiffusion and finetuned a custom model and integrated it into Discord. Lexica implemented StableDiffusion and finetuned their model with their own simple to use GUI.
StableDiffusion in itself is just command line tool but everyone is free to build their own GUI on top. Automatic1111 allows for the most comprehensive tooling but it can be overwhelming for some.
My trick is to keep a notepad which relevant info about the model and LORA's to achieve what I want c:
@@RinGoGuntheR Install the Kohya additional networks extension bro. You will get a metadata tab where you can store all kinds of info like "name, description, trigger words, cover photo" for your LORA
I swear being in the 3d field before ai made me accept these new ai models with open arms.
I don't blame artists that haven't established themselves being worried for their future, those who are already established can easily position themselves to benefit from AI. But someone who's dreaming of becoming an artist but hasn't even finished schooling themselves yet? That's a hard position to be in.
I don't think they will be obsolete before they're able to get a job in it, but what the job looks like will be changing as they are training. A bit like game developers since, well, forever.
Cant wait to generate a 3d city in one click
@@speedeespeedboi9527 Same, I painstakingly took the process to learn how to model characters.. So if I can take a tag team for an environment generation? JUST for the sake of rendering or animating things..? Please... there's so much that went into making just the person, let alone an entire environment...
There's a certain amount of irony with Adobe going full into AI now. For years the same digital artists crying about AI have based their career around Adobe tools without ever accepting that these tools themselves are the tools that put legit artists out of business in the 90s.
By the performance of firefly shown in the video, Adobe is gonna take a huge cut of that cake.
Personally I love the text to vector... for animations, will be awesome if its as good as they show.
I don't use SD, but I would use this instead.
this whole mess could have been avoided if it started off with royalty free images, and an opt-in. that would guarantee it remained open source and legal. and as Adobe has shown, just as high quality.
What "mess"?
@@lime148 Artists / designers revolting against AI. They want this technology to dissapear basically as it threatened their craft
Can you do a tutorial about modelscope text to video in google colab because the original google colab of modelscope is not working
I'm wondering whether people will find it "ethical" that they trained it on Adobe Stock images. While stable diffusion was trained on copyrighted material that was being shared around for free, Adobe's AI has been trained on images that people were trying to sell on their platform to make a living from, effectively invalidating their paid work with a little Terms of Service/EULA statement that everyone agrees to without reading.
"Ethical" is not "Legal". Adobe (and any other big companies, sadly) only cares about "Legal".
On a side note, I think the ethical problem is not the model training, but capitalism itself. Everyone would be happy to share their work(or art style) with the rest of humanity if they didn't have to worry about having enough money to live later.
Absolutely based and correct statements all throughout this video.
Yeah, well, how good is it for N S F W ?
More seriously, you could use Firefly to create image and feed them into stable diffusion training models. In fact, the entire copyright system seems pointless at this point.
This title... I can think of a million things that are TERRIFYING and Adobe Firefly is not on that list, and I'm a professional artist.
Idunno, I don't like firefly, mid journey is much more mind blowing. The way Ai generates images reminds me of how the brain generates hallucinations on acid. Very random unique and creative. It's perfect for generating reference images for ACTUAL artist not the average Joe trying to take short cuts.
The anti-AI Karen’s forgot that they are all DIGITAL artists. The digital realm is constantly evolving and, if they don’t embrace that reality, they will quickly find themselves irrelevant
How many "anti-AI Karens" you really know? You are the problem causing a divide.
Who cares what you think anyway. Just keep typing prompts you can’t copyright.
This is similar to Cmdr2's interface, but a lot slicker and a couple extra toys. like the text stuff, vector generation and instantly add illustrations to layouts. Lots of promise, but a very mixed track record. The "legal" dataset is going to be the HUGE draw for professionals who can't risk even a whisper of licensing lawsuit dangers, and the convenience of having it in the workflow and interconnectivity with other adobe products. The question is going to be will you get LORAs Textual Inversion and custom models?
Dude the only artists that pressed charges were the artists hired by the Big Guys. No artist in his own mind would press charges on somebody mimiking his style. It is flattering and also it created publicity. Latent diffusion is uncapable of copy pasting.
I have to admit, I really hated Adobe's business practice, but this Firefly AI looks so revolutionary. I feel conflicted.
It's not revolutionary, it's just combining functionality that was already in other text2img generators. Now adobe gets to throttle access to the technology and corner the market them.
SD is the Linux of image generators. Powered by a large open source community. It's not Windows. But it offers more control and creativity.
Exactly . Open Source wins
3D modeling might become obsolete in the near future. If AI can generate all of your models, videos, and animations, then there would be no need for drawing or even 3D modeling.
as a 3d modeler, I can only hope. I can't even imagine a day when flipped normals or a bad topology don't haunt my every waking moment.
The irony of this argument when stability has already worth millions of dollars while there is still artist out there starving to meet ends meet.
I don't like Adobe as company. But at least they respect the artist work and compensate them.
I saw the lemon character
Last call for Ai-cohol fellow Artists, now there are no more excuses Adobe just handed you on a silver plater the silver Ai spoon to munch on the creative juices, or are you gonna still hide behind the "NoAI" mantra and cling on to it like a baby to its pacifier.
But I use controlnet to combine multiple pictures all the time? Now with comphyUI its so much easier.
What is comphyUI? I am having trouble finding info on it.
@@NickEnlowe its a node based extension that is basically blueprints from Unreal. Allows user to determine workflow, so things like telling stable diffusion to first make a galaxy that transitions into sky that transitions into forest that again transitions into a field of flowers as foreground then can have it create individual or multiple characters playing in the field of flowers. Then have it blend and upscale the final product. All this in blueprint format so users can share workflows simplifying the learning process.
Why compensate, if they sold all right after submitting to stock?
When you talked about the people having concerns bacause of moraly reasons, it reminded me of the "Red Flag" that was mandatory for Cars a short period of time.
The Manufacturers of Carriages didn't like the "new technology" of horseless Transport, claiming it would be a thread to everyone because the people will be overrun on the streets all the time by the new "automobiles".
After that there was a time, every driver needed to hire someone who walked in front of his car, waving a red flag to warn everyone about the "dangerous machine" coming close.
Well... I guess with every jump in technology, there will be people having "moral concerns".
Open source community could create a free, sustainable and ethically sound dataset and a model, but seems like most folks seem to be like kids in the candy store, leeching images from ArtStation for training their dreambooths and LoRAs or using these models simply to replicate latest digital art styles or use those models to churn out pneumatic waifus. So better not be too critical of big bad corporations.
Yea this has been my biggest gripe. I'm in the Civitai discord trying to learn how train high quality photorealistic LoRAs but it seems like everyone is interested in training nsfw / celebrities / anime waifu's. Super unproductive degeneracy
I often have complex compositions, several figures with mutual action and background, and the AI is not particularly strong in this yet, and more often it is faster for me to do from scratch than to persuade the machine to do what is needed
But yes, for simple art in the style of Instagram photos - this is an outstanding achievement
don't take it too close, it's just that while this technology doesn't really help in my work
I'm interested in text-to-vector, haven't seen something that good on the community, just B&W vectors... that's the only thing is attractive to me, anything else as you say, already got it on SD
Here are my 2 cents Aitrepreneur.
These tools are not gonna make you an artist, or at least not the way I interpret the meaning of the concept of "art"
These are tools, tools to give you the ability to create pretty images, which is fine, but that doesn't necessarily make those results art, at least not to me which is the opinion I actually care about at the end of the day.
If your interpretation of art is just aesthetically pleasing pictures, sure, I guess this will give you the ability to make it and a bit more.
For me art is also a tool, a tool I use to relate myself with the information I received form the outside world, to interpret said information and be able to communicate back.
AI in the fields of art does not take away that ability from me.
I do admit that as an illustrator myself it will take away rather than give possibilities for new jobs (for what I've seen, the niche I'm in and how most clients don't give a damn) but this 2 things are not equivalent, at least not to me. Its what it is; is the result of being part of a flawed predatory capitalistic society, and the results of technological progress.
So reading a couple of comments here saying it would give the ability to everyone to automatically/instantly make art.....that's the whole point, it wont make you an artist, not just magically, or at the very least not a good one.
Noise is still noise, even with pretty makeup on, but I digress.
Just to clarify; I like AI and AI tools implemented in the fields of art
I do like that I would give tools/possibilities to everyone to try and express themselves in whatever way AI allow. Possibilities they never consider before for whatever reason. This would give path to new ways for art to be, and that is always a good thing.
After testing it out. Artist are Semi safe. It uses Copyright free images so no Celebrities, no Anime. Nothing Copyrighted.
14:20 actually this is good for companies. The problem with stuff like stable diffusion was that it was a legal mine field. Sure, you could use it without ever fearing getting sued but stuff like Disney and Warner couldn't without being dragged to court. Adobe might have won the AI art race.
The fact that there is nothing we haven't seen before doesn't concern me at all. Don't know why you're clinging to it so much. I wouldn't call myself an 'image generation enthusiast', but I like tinkering with stuff. One problem I always had with previous solutions is a lack of control over final result. Yes, can do everything they're showing here, but you have to jump through hoops to get what you want exactly - switching models, painting masks, using janky collages as a seed, iterating hundreds of times etc. The fact that you now can do the same but much quicker and without all this jank, is BIG. What Adobe did here is awesome (them paywalling stuff and making everything into a closed garden is another question). I really hope that this trickles down to open-source projects, and all these tools will become more consistent, streamlined and powerful. When I want to tinker, I'm free to do so, but when I'm trying to do something, I want to get things done without all this janky mess.
The whole Anti-AI Crowd is now embracing Adobe Firefly, because it's for "professional artists and an industry standard". SD and MJ are for "wanna-be artists" nerds. Firefly images are 100x better because they're "created by paying members" AND are ethical because of that.
Major drama when GettyImages goes to court against Adobe and Shutterstock, who has also partnered recently with OpenAI and will sell AI images. All built on the base of SD.
Having used photoshop from the beginning, this is insane
I think it does kinda suck that it's so mainstream because no artists can't even compete anymore with AI art and with the programs all being free Art as a profession is essentially dead. No more money in art anymore, because anybody can just spend $10 and borrow a GPU to make all the art they want.
I'd be happy to have a good tool that doesn't require hours of grief to get working.... Stable Diffusion I'm looking at you!
Can’t wait for Artstation to roll back their “created with AI” tagging mandate since every artists will quickly adopt AI in their workflow. If any pitchfork-and-torches anti-AI artist is caught using any AI features, I hope they get shamed into the dirt. All these people were so shortsighted that they couldn’t even see literal weeks into the inevitable future
Very well articulated Sir
@@Thesamurai1999 There is definintely a small but very loud sub-set of artists within that group that wanted to see AI be destroyed completely. Even in the most ethical sourced models they saw problems in that it would destroy and devalue the creation of art. They wanted all models to be removed from the internet and government to regulate hard on it.. even the ethically trained models.
@@eyoo369 Well, I don’t agree with those artists. And the professionals I followed were at-least in support of AI and saying it’ll be used in the pipeline. They were just against the bias some of the AI companies had at the time such as StabilityAI saying it’s ok to train their AI on copyrighted images while simultaneously saying training an AI on copyrighted music could lead to lawsuits. Or how Midjourney were fine with training their AI on small artists IPs but tried to avoid the IPs of bigger corporations such as Disney.
Nah, I'm pretty sure they could see this coming from miles away. I find it hilarious when you folks criticize artists for not wanting to learn to use these tools when you literally do the exact same thing by refusing to learn how to draw, perspective, color, etc. That same set of knowledge and tools that has been getting taught and passed down for generations and accessible to almost anyone who's willing to put in the time and practice. Artists aren't as stupid as you think. It actually takes an incredible amount of thought and know-how to create impressive looking stuff from nothing on a blank page. AI can be a blessing for people with disabilities. But I think what a lot of folks don't consider is the negative side effects of instant gratification it also provides. If society becomes so complacent and impatient from high level creative work being delivered in seconds and minutes, placing a high value on "I want it right now, now, now"...at what point does society decide that teaching and learning fundamental art skills is a waste of time/money and for people to just abandon it in favor of AI? How much practical long time knowledge would be abandoned in the name of technological progress? And with the threat of future wars, cyber attacks, EMP attacks, and power grid failures, there's risk of entering a new dark age. For example, if the power goes out, how do you use AI then? And how do you tell if a historical event really happened the way it's told if most of the images of that event were spawned from AI in an effort to fool people who don't know any better or any ability to discern the difference? (Example, fake images of Donald Trump being arrested.) This AI stuff goes way beyond just artwork.
I'm fully capable of accepting AI in the world, but we can't just go on pretending that there's no value in the fundamental knowledge that traditional and digital artists have been practicing forever. As far as art goes, someone who possesses both practical knowledge and skill and combines it with a knowledge of AI tools...well, that person is going to potentially be a badass.
I doubt (or at least hope not) laws will attempt to fully ban these AI solutions, it would just harm the country that makes the law as there will *always* be countries that are perfectly fine with immoral solutions if it benefits the leaders of said country.
I think the best outcome would be if the code and training material has to be made available for oversight, in such a way that open source AI don't have to jump through any extra hoops as it's already available. That would allow some oversight of AI companies, without hampering open source implementations. End result would be that companies that use xenophobic, racist, or banned training materials would have a harder time to get away with it (though it would obviously not rule it out) and as open source will remain a strong option you'd democratize AI and the beneficiaries of it to some extent.
Don't particularly think it'll happen anytime soon though, either the antediluvians that write laws will be tricked into banning AI, or like in EU put roadblocks for small companies to train AI, or some other luddite option that I haven't thought of.
How do I animate an AI using stable diffusion on Google colab
I totally agree. I do not see any revolution in the new Adobe product. I could replicate the features they showed 6 months ago using Midjourney, Stable Diffusion, or Runway ML. Instead of developing AI cloud-based innovative tools, they invested in Figma. As someone who used Adobe products for many years, I am disappointed.
Got an invite today. Some data definitely trained on Adobe Stock, bcs you get near stock style photos)
you know the history of AlphaZero ? that Go board game AI that started from actual zero data, the developer deliberately make the AI fight the prior released AI, the AlphaGo which is trained from millions of real human Go professional players gameplay, until AlphaZero managed to win continuously in a row, and yes they did it, AlphaZero is smarter than AlphaGo while only fighting the prior AI to get the data.....imagine if MidJourney used Adobe Firefly to create MidJourney-Zero
I've uploaded things to Adobe stock.. so they definitely have images trained on any old shit including licensed images right?
Now the promt master will protest againts adobe firefly
These AI art tools will help people to make something creative.
how can they compensate artists if their image was used in generations, there's no training data in the model file, it's literally impossible to look at used weight values and be like, yep, this weight came from this reference image. The only thing i can guess is they keep a copy of the training data on-hand and if the user types in a keyword, they lookup all the training data that matches that keyword even though some or most of it of it may not be in the actual image. Their only doing this to make the public outcry happy and to play into the mis-understanding that stable diffusion is a collage tool. In other words, if you like to think it's a collage tool, they'll just let you keep on thinking that and profit from it.
What I am interested in Firefly open source or at least modular like will there be a way to load a different SD model into there datasets or tools or use embeddings etc. If so you have the best of both worlds the copyright free image dataset and the ability to use other models etc. That would protect Adobe from any lawsuits as they are not providing any copyright images. By the way I'm not saying other models are illegal I'm just saying Adobes model is "legally safer" other models could at some point cause problems in court etc. I'm hoping that using images to train does not break copyright but you never know how the courts will rule.
The problem is that people rely on art as a source of income, and without it they can't survive. If we lived in a world where everyone had access to basic provisions and could do what they really wanted to do, regardless of its profitability, then artists wouldn't have to worry about being replaced or left behind economically. Art is not a necessity, just like video games. What you're really flamming is that capitalism forces us to be as efficient as possible, creating monopolies and making it impossible for small developers to make art and games that aren't profitable. The only valid criticism left for them is the unconsensual stealing of copyrighted art.
Seems like they're trying to ''Apple" this. All the things we've had already in a fancy ui. Normies will think Adobe invented it all.
Obviously revolutionary for Adobe fans
work local or online?
There is already existing model for text to template. I can't remember it's name. But it exists. So in essence they made three models : text to vector and text style and midjourney blend. Everything else can be replicated in automatic.
The difference will be in the integration with existing Adobe products
Ease of use is always a selling point. Which is fine as long as open source is available.
In a way, AI is simply ease-of-use as you *could* do it all in open source programs even before SD/etc.
Adobe is going to nickel and dime their customers to death when Firefly comes out of beta, just like they always have. They're not going to care that this technology further widens the gulf between those who can pay their licensing fees and those who can't, regardless of the fact that Adobe tools are the _de facto_ standard among professionals. Still, I think we can agree that it's better that this alternative exists than not.
My worst nightmare would be if Adobe buys Automatic1111
prompt is never going to be the future, there will be no "prompt engineer" job as you think, its just a way for us to use & understand this early stage of machine learning technology . It will stay here for a while but it will eventually replaced into more fluid & integrated workflow, this is one of the examples, more controls, a ways to edit & do some revision on the arts.
The tools we use are 'absolutely free' because we arrogate to ourselves the right to think that an image that ended up online is automatically in the public domain. And we train our models.
This is convenient, but not really correct, especially if the images created are used not for personal enjoyment but to make money.
You get old memes! You get old memes! ..old memes!
I don't see the problem with AI using internet data to train models.
It's on the internet. It's now in public domain.
The art was never stolen and the images that AI make are original generations.
In fact, its impossible to even replicate any actual drawing using AI alone.
It's not like google or chat where you type something in a poem and it gives you the line by line text.
AI art is trained to generate new images using their models as references, aka guides.
So if you type in that you want the mona lisa. The AI will give a memory depiction of the mona lisa, except without any specific color, subject, paint style.
It would give you a random woman in the same pose as the mona lisa. Thats far from saying the mona lisa was stolen.
Last time I checked, in the art community, referencing a model is used by everyone. But when AI does it, its unethical?
Also, if AI has to pay people now, then so does Google for all the illegal images they have too without artist compensation.