The Rise of AI Generated Art

[@bycloudAI]

Second time making a semi-documentary, do you enjoy these kinds of video style? Please let me know! And please stop flaming how I pronounced DALL-E I swear the extra E stands for my shitty English 😭 And of course, the first 1,000 people to use the link will get a 1 month free trial of Skillshare: skl.sh/bycloud06221

[@eduardofernandez2469]

Hello. Maybe its possible to convert 2d images in 3d. Like cartoons, animes and 2d games

[@prest01]

where is he

[@eduardofernandez2469]

@@prest01 what ?

[@3333218]

I sincerely enjoy anything you make!

[@SlavicAMV]

yes i love these documentaries, the editing and everything makes it so smooth and interesting!

[@nilsqvis4337]

"Choose a creative career", they said. "It'll be among the last to be automated", they said.

[@hombacom]

No matter how much we improve the tools you still need to be creative to make something that stands out

[@zubinkynto]

@@hombacom until the tool becomes more creative than you

[@hombacom]

@@zubinkynto that doesn’t help if it not understand what you write

[@asdfgh-ym5se]

Tools don't need to understand to replace humans. They only need to produce output that compares to or outperforms humans who do understand.

[@hombacom]

@@asdfgh-ym5se creative people with all available tools always outperforms non creative people

[@claradrake]

As an artist I'm so damn scared for the future...

[@s-zz]

The enjoyment of creating artwork, is something a machine can never take away from you. Assuming you enjoy the process, that is.

[@Y0UT0PIA]

@@s-zz "The enjoyment of sowing the fields is something no machine can take away from you, farmer-kun" oh yea, gee. thanks bro.

[@apeDLuffu]

@@Y0UT0PIA LOL

[@hombacom]

@@Y0UT0PIA Cooking food is still a creative work

[@jacobsz7465]

here is how you can use AI as more than just an art generating tool but simply a tool to aid the process in making better art kzbin.info/www/bejne/nIS9gpZsnJ2Eadk

[@Hhhh22222-w]

For artists, the biggest issue with AI generated art is not how the art itself, looks, but the time it takes to the AI is generated what the user wants, if the user doesn't quite like the art, they can just run it again until they find one they like, you can't do this with artists who will take time.

[@jacobsmith1877]

That's why I think AI art should be used for rapid prototyping and concept development, then have an actual artist create the final (if you are doing professional media)

[@p1roncho]

but in order to get good AI results the client has to know what they want...so we're safe lol

[@seanrodrigues8184]

Umm yeah. That IS how computers work Tanto.

[@youtubedeletedmyaccountlma2263]

@@p1roncho wait for us programmers to modify the software. We aren’t dumb

[@someuser4166]

@@youtubedeletedmyaccountlma2263 you mean wait for github copilot to modify it

[@uglycasanova6689]

This is going to put a lot of people out of work. Just messed with midjourney for hours and it's just mind blowing the stuff it creates in mere seconds.

[@macjonald]

Yeah it’s awesome that these scientists picked a very pressing issue to tackle with AI. World hunger? Nah. Climate Change? Nah. Global Conflict? Nah. People Making Art to Pay Their Bills? Yeah, fuck those guys.

[@xxmemestar69xx82]

🤷‍♂️

[@Memememe-is1yn]

I am a novelist. I hire artists from around the world around $300 per job to make covers for me on Fiverr. AI is about to destroy their jobs.

[@sarasawicka]

Soon AI will write novels and you will also be out of job

[@Memememe-is1yn]

@@sarasawicka Very possible. I guess I'll just have to get all of my ideas out there sooner rather than later.

[@ursmeyer5913]

@@sarasawicka exactly what will happen. AI will write 4‘000 Novels a Second.

[@magostadepaodequeijo6325]

You will still need to hire someone, but instead of a an artist, you will need to hire an editor to make the image count as "manmade" as purely AI made "art" can't be copyrighted, it's the same as picking up a random stock image you pick up on Google and put it on the cover

[@kikc]

Aren't you happy that you will get your covers faster and cheaper? What is wrong with you.

[@Y0UT0PIA]

As an artist, I wanna give some insight into the markets and what I think the effect of this tech is gonna be. First, I think it's important to distinguish between freelance and working in the industry proper. Freelance-type jobs generally aim at the creation of single illustrations - Logos, banners and the like, stylized portraits, character art, book covers, these are areas where artists will very soon be completely outclassed by AI-based image generation. It's cheap, it's fast, it's always available, it won't ghost you because it's having real life problems. If you didn't formulate your prompt right you'll get as many redraws as you want, and it can mimic any style rather than being limited to a few approaches it has specialized for. And since you're asking for individual images and aren't looking for a super specific result, the fact that there's currently quite a bit of stylistic variance between images doesn't matter as much. Now, by comparison if you're making assets for even something as simple and silly as a mobile game, the results you're looking for are going to be much more specific, and an AI isn't going to design you a cohesive user interface, let alone game assets that share the same style. And this is even more true for more comics, movies or more artistic game design - since you used hollowknight's OST for this video, let's just take that as an example. Such a game's aesthetic is intricately structured - you have a general aesthetic, within that different 'themes' for the various zones and recurring design elements that tie things together. Enemies are designed to have conceptual interest, so that things gradually build up from the 'common' to the 'unusual' - from regular ol' bugs to bug nobles, mages, preying mantises and so on. In a word, this kind of project is like a visual symphony, where the various elements are all designed in relation to each other, as part of a cohesive vision for a whole. This kind of thing requires 'persistent intention', which is what generation lacks. I think AI is going to severely impact 'art as craft', so to speak, which is basically all of freelance and at least a good number of industry jobs (people who, say, paint unimportant background assets andte like), and at the same time what we have now isn't hugely useful for actual capital-c Creatives, since it doesn't give you enough control to integrate it into a coherent workflow. I'm looking forward to genuinely functional AI-based solutions to simple but time-consuming tasks like 'turning a precise but rough-looking sketch into clean, finished lineart' or 'rendering out an image based on some key lighting- and material information', ideally integrated into painting software directly. Anything that allows people to speed up their workflow without losing control over the kinds of results they end up with would allow individuals and small teams to tackle much more ambitious projects, and will result in a new golden age of indie animation and games production.

[@EudaderurScheiss]

as a game dev with a programmers brain I can completely agree. it is not that AI will create everything with one click, it will be used as an asset to iterate over tons of ideas. the artist still needs to combine those into a final image. i tried to design a game logo by my self. I failed and hired an artist. we iterated over ideas forth and back (this bigger, this smaller, this there, this other color etc etc) and to miscommunication the last step got fuped and i was really unhappy, even though the logo was good to ok. two days ago i found an AI that can generate logos. i dropped in what i had and it just shows me alterations of which i really like a few. it gave me ideas and parts which i will combine to a final design. (i can do designs, but it just takes me just too long). if the artist would have used this tool, he would have saved 5 ours+ of initial work, presenting ton of basic ideas and finalize it 10x faster. and the annoying part of the initial iterations would have been enjoyable. oh and I know something to draw into line art, i used it to digitalize my girlfriends handrawings and to vectorize it... I'm not sure if good enough for your needs: sketch.esslab.jp/ Are you working in the game industry? There is so much cool AI stuff coming

[@gaboqv]

Not to break your bubble but you can easily retrain the network to follow an arististic style, altough at the moment we don't have a very good way to do this, I bet in ten years this would be very easy

[@random__handle]

@@gaboqv ^this

[@yiannchrst]

Yeah, i totally agree

[@richardhernandez1101]

@@gaboqv If you claim something you need to provide evidence saying ''retrain the network to follow an arististic style, altough at the moment we don't have a very good way to do this, I bet in ten years this would be very easy''. This just makes you sound like one of those people who believe you're right about everything.

[@bibirmengkeroet6717]

10 years ago when i was a kid i have a dream working as artist in the future, but my mom doesn't allow it and says "You'll never get a job". Now i know why. Thank you mom

[@patriciaszabo8015]

Art therapist will be needed. Mental health issues is going to rise somewhy.

[@AmazingArends]

I think this video got more technical than most people were expecting. The title was “The current state of AI generated art,” NOT “The technical details of how AI generated art was developed.”

[@UsaraDark]

That stuff at the beginning was the stuff you see while high on shrooms. Some really intense stuff, especially when it envelops your entire vision and there's no escape. Put someone in VR with those images with no escape and it's a lot to take in.

[@SephTunes]

Nah. It's much more detailed while tripping. Like, endlessly more so. Cannot be expressed

[@Mede_N]

Awesome video! May I point out one possible aspect for improvement? Sometimes the music track is too loud and drowns your voice, so it's harder to understand you (e.g., 17:25, the piano frequencies are in the spectrum of (female) voices and therefore human hearing is quite sensible there and thus the piano music "collides" with your voice). Cheers!

[@joejose8433]

That's because everybody tried to be a music f****** producer when they uploaded KZbin there's something in the most editing software called Auto duck it means that when you speak when the music is playing the music is automatically lowered now I don't f*** with Adobe because it requires a license I use audacity which doesn't require a license and is open and source freeway so I know when I speak the music is automatically Lowered so look it up #Autoduck

[@daedalus_00]

I have a prediction that I haven't really seen talked about. I feel like there will soon be a new art skill, the skill to 'prompt' the AI to get intended results, and a wide variety of prompting tools will be developed. This video mostly showed text base prompts, but nVidia has an AI image generator that is prompted on label masks. Essentially, you color in if you want water or a house or trees or whatever, and that part of the image becomes whatever label you painted on to it. Now imagine a tool that you start by telling it what you want. then maybe you select images you like and it will regenerate images based on what you like. Then you can further alter things with pose tools, and painting labels onto the image. Then maybe you fine tune even further by adjusting sliders for art style or other parameters. Also, I wouldn't be too scared of non-artist 'taking over' the art world, because the unfortunate thing is some people will never have a strong imagination, even with awesome AI art tools.

[@lizbotnick]

I think it'll really create added value to becoming learned regarding art history, art styles, different mediums, and individual artists.

[@tokyofamily8536]

this feature could be merged with google images, just like how when you search for an image and click on it, it displays similar ones

[@darksunrise957]

@@lizbotnick Oh, wow. This reminds me of how a number of sci-fi settings treat technology as some sort of god, magic, or higher power, and have the equivalent of priests (I guess the most literal version being Warhammer's Tech Priests) who don't so much manually control tech, so much as they know how to influence it in the right way to get it to accomplish something. No more buttons and levers, but communicating in a strange, inhuman way to make technology perform the tasks we humans want/need. Not to mention, sometimes tech in those settings can be fickle and very choosy, which is very much the case with these AIs today.

[@yorch802]

I disagree, the people creating the tools have better understanding than noobs that just wanna prompt them.

[@eurasiaacaci.-110]

Prompt ai can do that.

[@CG64Mushro0m]

6:24 the big sleep ai is honestly one of my favorite txt2img ai's because of its weird... dreamlike? ability to generate images that dont fully make sense, and yet, you know whats in the image, even if it looks distorted.

[@krysidian]

Fantastic video breaking down the essentials when it comes to image generation. Really fun to see the history of it all, how these AIs relate to and basically created eachother and simply how fast this area of AI has evolved in such a short amount of time!

[@BeekersSqueakers]

The irony of a SkillShare ad on a video hailing the destruction of the creative skills market.

[@yolocrayolo1134]

when you test the ia and realize they are still unable to render green clothes or well formed cartoon faces

[@flynntaggart7216]

We aren't destroying it instead we are making our own new competetor

[@mizomint4197]

Just wait. We'll have an AI generated movie with AI generated invented actors and actresses in no time at all.

[@ursmeyer5913]

The Movie is not the Problem. Mass Media will be.

[@kikc]

That will be so good

[@kikc]

@Cassowary Egg Having an infinite amount of films instead of spending 20 minutes searching for something good to watch is better. Not having to wait between the release of sequels is great. Having *good* films is great. I don't understand why it would be so bad that you'd feel the need to insult me

[@forfreedomssake4315]

It's gorgeous art, but the speed of progress of AI is just terrifying...

[@mityashabat]

My favorite AI channel, love your videos, I hope you get more recognition

[@BrassPetals3Voices]

I’m finding with Midjourney that the fewer words I use, the more ‘creative’ the finished piece looks. If I’m aiming for recreating objects in Blender as models, the more words, the better. Fewer words tend to generate images heavy on a scarlet, black and white palette while more words generates a more diverse palette. One thing I have especially noted is that it is great for matching/generating mood prompts with image. I haven’t seen a mismatch yet.

[@muleboy3537]

How are you able to access and use this. The rest of your comment is just flexing this privilege.

[@Veptis]

Thank you for the overview and all references. I am working on my own "text to image" ideas that's more like a language model fine tuned on shader code... But the explosion of results in the last half a year really force me to write about it.

[@LihimSidhe]

So... I was inspired by your video. After realizing that Imagen wasn't released yet and Midjourney/Dalle were invites only, I then tried to use 'Centipede Diffusion'. It's honestly very confusing and it's hard for me to make heads or tails of it. The tutorials out there on how to go about this are VERY limited and assume that a viewer brand new to this world knows a lot (ex: google collab notebook). I have an illustration and web design degree and I STILL was very, very lost. The furthest I've gotten is finding the default Disco Diffusion notebook, clicking 'Run All' and now I have a bunch of lighthouse artworks in my Google Drive. Is there ANY chance you'd be willing to do a 'Using AI collab notebooks for Dummies'?

[@skylordianandy2644]

At this point I'd just learn to draw if I were you.

[@Graphomite]

This tech is going to begin like the calculator is to math. It's going to be the sidekick that every artist will utilize. This will be exciting at first, though, it seems very possible--if not inevitable--to create a piece of software that simply provides more efficient, consistent results than human beings, demoting the whole medium of handcrafted art to a largely unprofitable hobby, like crocheting or woodcraft. Arguably even less relevant. In the eventual world where every digital or printed image must be assumed AI generated until proven otherwise, how will handcrafted art be advertised? Every piece of art would have to include preliminary images of the unfinished product. Say goodbye to images impressing people at a glance. There's gonna have to be an Etsy for "real art" once AI perfects the craft. It seems strange to associate the first step of the AI revolution with an art generator. It's not exactly what anyone expected to be the harbinger of singularity, and yet, it *is* the first example of generative AI wheedling itself into layman activities. Flawlessly so. The fact that it's meddling with art of all things is almost disarmingly impressive. It's so surreal that it's hard to fathom a world with it. And in a few years, it'll be hard to fathom the world before it. Your children will not see image creation as an artistic labor. Our pens, pencils, and stylus will be to visual media what typewriters are to the novel. Impractically laborious and irrelevant, since nobody would be able to tell you used one, anyway. ...except of course the content of a novel isn't reliant on the process of its printing. Although I can imagine an alternative universe where calligraphy and typography was a major and widely practiced and accepted part of novels before PC devolved the art into "simply jotting ideas down." In such a world, modern novel writing could very well be considered a lazy, passive artform. Kinda puts into perspective how easily we'll be able to normalize "original art" as being mostly promp-typing and some personable fine tuning.

[@LarsRichterMedia]

Also say goodbye to advertising yourself online as a digital artist. All your work that you upload to some public space to attract people's attention is going to be harvested by a horde of different AI systems that will contiously scan the whole damn internet in a race of "who can collect, process/cleanup and feed the most amount of data to their AI" - You will have to almost completely hide your work behind an opaque watermark, which is going to help attract a lot of people for sure. Since these AI's also feed on tiny thumbnails you couldn't shrink your images enough to protect them from AI harvesting. The tech itself is mindblowing, the world it begins to paint for me looks absolutely terrifying.

[@aegisreflector1239]

​@@LarsRichterMedia Agreed. A.i the new/old God

[@NewtrendsMe]

@@LarsRichterMedia Not the typical "Skynet" everyone had in mind, huh? It's like a lovecraftian dystopian novel. Written by a bot of course.

[@LarsRichterMedia]

@@NewtrendsMe and it is only the beginning.

[@Vigil314]

Just think of how google's deep mind AI completely shits on grandmasters in Chess after only a short time. Now apply that to a mature and fully developed AI like midjourney (not even it's final form) pairing deep learning with art generation. It is my pessimistic opinion that most artists; 60+%, are going to be out of work, and lining up at mc Donald's for a job. (me)

[@desu38]

One issue I see with these text-to-image models is that it can only make what you can express in words. That's not an issue I have with something like Artbreeder or Looking Glass.

[@DRIVECLUBistimelessPS4]

Not necessarily, you can add image (url) prompts in Midjourney too

[@ShaneSemler]

Excellent video. I just discovered Disco Diffusion. I've been using Wombo Dream to generate ideas for paintings, which it's pretty good for. Stuff like Midjourney and DD seem to be aimed for more coherent images, which isn't always desirable. Anyway, I'm sure alot of illustrators are going to be out of work. I'm glad I'm not a professional illustrator any longer.

[@Mosi_O]

great intro 0:58

[@RuneViniGarcia]

Holy shit amazing content, happy to have funded your channel

[@MyCygnusX1]

As an artist I am excited to use these AI tools to sketch out my initial ideas, and then create something that looks nothing like it! Thanks for these essay style vids :)

[@macjonald]

Illustration will be devalued to the point where no one will pay you, thanks to this crap

[@flynntaggart7216]

@@macjonald why i will pay for a simplest thing like a beautiful wallpaper a thumbnail and banners for yt

[@Morita139]

What I'm fearing is that if all jobs could possibly be taken over by Ai, and there will only consumers left, but said consumers without a job, how will things work?

[@flynntaggart7216]

As a programmer y e s every job can be automated and optimised by ai so it can perform better than us but this will not happen do quickly so stop overthinking

[@Ryuuko3]

Ai work for us

[@flynntaggart7216]

@@Ryuuko3 We will see in future who worked under who

[@users4007]

luxury communism

[@flynntaggart7216]

@@users4007 cringe

[@huskaibeats]

genuinely one of the best videos with very in depth detail, you did a very good job looking into the discords finding the info :P . thank you for crediting everyone aswell!!!!

[@gitomiko]

This is the most helpful video about the state of AI today, that you will find on the internet. Thank you so much 🍻

[@gingercyclops4397]

Programmers are now better at art then all artists combined, science ftw yet again

[@FinisterreTV]

its crazy how fast this is being developed and improved. singularity soon

[@kato_dsrdr]

Sheesh, the fact that computers went from solving math problems to creating art in just a few years is just crazy.

[@SMmania123]

Disco Diffusion, DALL-E 2, and MidJourney were all so impressive!

[@CastleMiser]

I know another one called nightcafe which is just midjourney but a customizable but crappy

[@powers-aj]

nice editing fr

[@Kram1032]

Very cool video all around. Surprised you didn't mention Imagen as that seems to be a step up over DALL-E 2, maybe not in visual quality (on a base level they seem about on par) but in terms of understanding and incorporating many more details of longer, rather complex prompts. (DALLE, DALLE 2, and DALLE Mega all tend to only pick out a few parts of longer prompts, ignoring other parts. It's not that Imagen *completely* avoids this but it seems to be much less of a problem)

[@LetterRip1]

Likely related to the attention model, Imagen uses T5 a much larger language model that has more attention blocks.

[@Berserq]

@@LetterRip1 Imagen is in a very closed beta. Very few people even get to try it.

[@dgspitzer4771]

Thank you so much Bycloud！This is so helpful～

[@previouslyon8807]

I was watching this video, and halfway through a stock music song I wrote started playing in the background, there are billions of videos, what are the odds of that happening?

[@Wander4P]

Um, actually there's only one 'E' in DALL-E 🤓 great video btw

[@s-zz]

Every day we get closer to the extinction of human artists and the rise of machine learning. It is both a fascinating and terrifying time to be alive. Regardless I'm very intrigued to see where it goes from here...

[@ShaneSemler]

Nah. Just because there's IKEA doesn't mean no one makes cabinets by hand. Take a look at the maker movement.

[@s-zz]

@@ShaneSemler Actually a really good point, only problem is. How many hand made chairs or tables do you have in your home?

[@ShaneSemler]

@@s-zz Touché. I don't have the money for custom furniture or the space to build it. But I would if I did. Also, small art pieces are more accessible than furniture, which was just an example.

[@chaosfire321]

@@ShaneSemler True, but the central issue of artisanal custom made stuff is that you can't really have that be the ENTIRE market. How many people prefer going to a special hole-in-the-corner-coffee shop vs Starbucks or Dunkin. Artistical stuff is niche by it's nature, which means a ton of competition catering to that niche.

[@leveledfeats]

I dont believe it will extinguish human artists. Sure Ai looks pretty cool, but it will only shine more light on artists who have made good art in the first place through hard work and practice. More influx of normal people joining on the AI art bandwagon --》the more it creates scarcity of human artists who still paint and draw with good fundamentals and they become more valuable as a result over time. People will begin to crave the quality, identity, and voice of an artist once everyone begins drowning in the Ai rabbit hole. Just like how modern gaming has declined in recent years as studios focused more on high end graphics than storytelling and game play, people miss the old games. Same principle applies here.

[@p_p]

i saw some captcha images visibly recognizable as made with ai

[@memegazer]

The only advantage big tech has on grassroots independent AI research is their data...democratize the data! LAION ai ftw ELEUTHER ai ftw Lets make the linux of ai a reality

[@rky.91]

Hollow knight ost spotted near minute 7, but with different pitch

[@mfpears]

The future of entertainment is going to be overwhelmingly amazing

[@shalipicks1234]

Incredible video. Incredible channel.

[@zeekmathis9801]

Great vid bro

[@shApYT]

Nut from Trees marketplaces? Also make a follow up on Imagen from google.

[@firosuke]

Excellent video. I wanted to share my emotional response, because it surprised me. I may not be the only one who feels this way; human artists will be pleased. For several minutes, like everybody else, these images blew my mind, one after another. This will change everything, I felt. Who can imagine what renaissance this will unleash in art and animation? I was even more mind-boggled that one day a future generation will see this technology as something completely normal, rather than an incredible revolution. But then something unexpected happened. I was gradually desensitised. All those garish, vivid hyper-realistic colours and clashing images, started to grate. It felt like an endless rainbow of beautiful candy, bead after bead. I wanted to stop consuming it and look away. I started craving the limits and imperfections of something recognisably human. I believe that when a computer does the majority of a task for us, we often don't work or think hard about it. If not, the result will be generic. Anyone could have pressed that button. It can't be important to you if you haven't worked hard on it. Others have said that the real value of art is that it lets us share, communicate, and re-experience our feelings of what it means to be alive - - our unique corner of the human experience, with joys and suffering that we are grateful when others recognise. Yes, generated images can evoke an unlimited variety of ideas and feelings, without a conscious creator. But that's exactly what makes it less appealing to me. I felt the importance of emotional connections to real people through art, whether they spent hours painting it, or perfecting prompts to express something unique and personal to them. Either way, we want to cheer and support them. I am not an artist, but all these images made me want to pick up a brush and make something for myself, as a human. I wonder if there could be a cultural backlash against this tide of imagery, like how Miles Davis' "Kind of Blue" paved a new direction in jazz, cooling it down in response to the fast-paced frenzy of bebop and 'hot' jazz.

[@RegularRegs]

Good video bro. Just want to point out that the word modal is pronounced like. Mow-dul. Rather than model. They mean very different things.

[@max477]

Which of these can we use commercially. I think Dall E2 is not available. I could you please suggest which of these can we use commercially.

[@iternityhuman1782]

@ 16:53 some of these AI pieces are really incredible,, many gr8 examples in your video.

[@jhay_vine5083]

This channel is underrated

[@basilunderworld]

i cant wait to watch a video called "the rise and fall of ai"

[@inxomnyaa]

After joining the Midjourney beta yesterday and trying some stuff, i found exact image in the thumbnail :o In general, the results were pretty amazing and the discord integration is pretty cool

[@yosha_ykt]

where is google Imagen?

[@someuser4166]

I've always been one for hardwork. But with ai taking the skill I've worked the hardest for in my entire life and just giving it to lazy people who couldn't bother committing or disciplining themselves to earn it I'm just gonna become a government sponge and leech of the tax payers. It seems hardwork doesn't pay so I'm done trying. Like Hank Williams said "if those rocky road don't start to get smooth I've traveled it long enough"

[@Kelly_Jane]

Midjourney is good, managed to refine me solid base art for my cover using just the free trial generations.

[@VaunaKiller]

This loud music when DALL-E 2 rolled in was really jarring. Great video anyways

[@ARTificialDreams]

Great Video! Very informative!

[@3drocket]

@bycloud Thanks for the video. This was way more detailed and interesting that I had expected when I clicked on the thumbnail. What would be really interesting to see is this AI art generation applied to 3d models so that you could describe an object and get some 3d models of the object back from the AI.

[@danyal3535]

Heavily underrated video

[@Danklovic]

Great work

[@karaokewaala9107]

This is soo good man.

[@cs6993]

When you take skill out of the equation

[@timrobins98]

RIP detail, Thanks KZbin compression

[@wetmote8221]

you reckon ai art is a threat to all artists out there?

[@iatoa2812]

Amazing video!

[@MiguelSilva-br9cb]

Am I the only one or right after the 3:10 or something mark the music is basically a variant of theme song of LUIGI'S MANSION ?

[@Mishn]

That's because it is! It's a piano cover, this one in fact: kzbin.info/www/bejne/rqrTlmt9hK2NpqM

[@bycloudAI]

lmao can't believe this video is already outdated in just 3 months

[@flynntaggart7216]

Yep

[@dpredie]

how to create the music video?

[@ryanmelvey8764]

imagine when it starts rendering 3d worlds...

[@flynntaggart7216]

Already started

[@kikc]

Hopefully soon

[@Megneous]

1. You should not have your intro music so loud. It seriously hurts those of us who use headphones. The difference between your normal speaking voice and intro music is too loud. 2. You apparently completely skip Imagen, Google's answer to DALL-E 2. 3. Please don't use static sound in the background of your videos. Those of us with sensory issues can't stand to listen to a video if there's popping and crackling in the background.

[@TonyAube]

How did you create the animation for the intro?

[@diogeneslr]

Incrível seu video. Sou brasileiro e gostaria de entrar nesse mundo da arte AI mas me parece uma bolha fechada por enquanto. Como começar? Por onde começar? Obrigado!

[@gabkiu7538]

recomendo esse vídeo, kzbin.info/www/bejne/bWGXk6WvrM6snNk , ele também explica como conseguir acesso e como utilizar

[@diogeneslr]

@@gabkiu7538 obrigado pelo apoio!

[@synthoelectro]

I think the original google eyes are still in all AI, because AI images look strange, no matter how much you work at it.

[@TomDenneyArt]

Nice Mic. Did anyone tell you not to swallow it?

[@jakethedog97]

the rifle is invented, the bravery is broken

[@yadishansar]

And now there is Google's "Parti" which looks better than both Imagen and "DALL-E 2"

[@Qu0thTheRaven]

Dall-E E-2 is pretty cool

[@soldesign72]

Imagine text to video

[@DonnieTheGuy]

The thing with artists and such is that for now, the AI can only give outputs based on its knowledge about the text input. So for things like branding for small businesses and artwork for projects, even if you use it purely to design something from the ground up with the help of the AI, you would pretty much still need an actual artist to do the job. Another issue that comes with this AI is that, unless it is trained to ignore/reject/censor prompts with immoral/illegal/inappropriate context, it can be used for evil if it ends up in the wrong hands, and it can cause a lot of confusion since some of the models are already 95% in pair with reality.

[@lognarki]

ignore/reject/censor prompts with immoral/illegal/inappropriate context - is censorship that kills technology. People who want to be offended by something will always find something to be offended by. OpenAI and Google are afraid that people will start making NSFW content? But people will produce it anyway. So, from their point of view, real NSFW content made with real people (or, God forbid, children) is better than synthetic content that no one was harmed in producing? I don't know what to say here... Will their networks be used for political fakes? They hid GPT-3, then GPT-6B/ GPT-J/ GPT-NeoX came out. How many fake news stories have they created? About zero. Fakes are written by people, hand-written, thoughtful and believable. Fake photos? Pfft, 99.9% of synthetic images have clearly visible artifacts . Even StyleGAN. It's believable enough for artwork, but absolutely obviously counterfeit for detailed photorealism. All OpenAI has achieved is that they have killed AI Dungeon with their censorship. And people left to generate NSFW fanfics in Holo AI on GPT-J. Because NSFW fanfics existed, exist, and will continue to exist. They're not hurting anyone. And the people who write them are better off continuing to write, read, and watch synthetic content than to go looking for a replacement for it in reality. And if they are all so ethical and highly moral - I don't see them banning Object Detection and other similar technologies for use in the military and defense industries, where such technologies literally kill people. Ofcourse, synthetic porn is certainly a much scarier thing these days than drones killing people with automatic targeting. The age of hypocrisy...

[@401uwu7]

Plz anyone help me...I have a video of my parents wedding from 1993... video look terrible...you can even count pixel...i want to upscale it to 720p.In video the face is so pixelated...i have seen one of your video the you scaled an old picture using ai...can i do this on video?

[@carolinewhite2958]

Hi guys, you can enhance & restore blurry faces in wedding videos following this video tutorial: kzbin.info/www/bejne/aJW5qKeurMaMsLc

[@nananina982]

Maybe you can extract each frame and scale each one individually

[@401uwu7]

@@nananina982 7000 frame that's a lot it will took ages...

[@kikc]

@@401uwu7 use the children in your basement

[@EudaderurScheiss]

love the video! one tip try to normalize the sounds and your speaking voice. i was like too loud.. oh not loud enough... ok too loud again

[@markwrede8878]

Stray cat in dark alley looks for marks in his confidence game while smoking a Marlboro.

[@manzell]

A lot of not that smart people made money off the early internet in the late 90s early 00s. I wonder what kind of simple ideas will prove to be big money makers in the early AI era?

[@Mrmcwarpather]

i think about it all day 😉

[@Eagles_Eye]

Wish there were more versions that you can use in web browser 😔

[@claradrake]

What about wombo dream, it's an app who generate images to text, and it seem it does the studing on the subject via internet ( like he google itself lol )

[@jaredf6205]

After using VQgan for over a year, I really dislike Disco Diffusion. I just can’t get anything I like out of it, I’ll try over and over and nothing comes out right or looks good. It also seems to only be interested in making single objects so making any kind of pattern just doesn’t work while in VQgan, you could get beautiful full res patterns after like 20 frames.

[@boigercat]

What type of Chinese tho? mandarin?

[@metakron]

In my opinion if it hadn't been for dalle 2 and chatGPT no one cared about AI

[@HenryPhD]

Bedankt.

[@whatever-fy6dk]

Now cover Imagen and Parti.

[@ItachiUchiha-tu7ir]

Can someone tell me which Anime is 18:01 from? My sharingan is not working..

[@someuser4166]

I think it's from Golden boy

[@crashdummyglory]

The video is good but i wish someone shed some light on what these apps do in the background. Meaning how do they source their images? How many images do they pull for a given text input? For examole Why is it that vq gan clip has a very similar output with the same text? Does it mean it uses the same image or images each time. This can be limiting. As one image has limited information like form, lighting, texture specific to that image. What r these apps drawing from?

[@itchyeyesmcrealize165]

The AI images make me feel weird ...

[@dalle2andme224]

As an A.I. Artist, i must say, has Digital Art replaced Physical art? Or has pouring moulds replaced hand chisel sculpture? If you are a talented digital artist with vision and flair you will do well. We are not coming for your jobs, in the art world process is very important.

[@AGM_V2]

AI artist, how long do you think till AI will be able to generate its own prompts using the data you provide, there is no future or coexistence with AI. Why the fuck would any one pay you or any other kind of artist when they can ask AI to generate what they want and if they don't like do it again till they do. Its this naive thinking that is going to be artists downfall, if you won't fight for your rights against these gready corps who steal and will profit from your work than you will be replaced, its not a tool its a cheep, efficiant, fast replacement that average artist cannot complete against. And if you think you will get paid for editing ai work or entering prompts you are highly mistaken ai will take that as well

[@donpaisa8627]

Don't fool yourself, you are not an "artist", you just write prompts. The skill required to do what you do is the same as mine writing this comment.

[@chaosfire321]

I've seen a number of haughty artists acting as though the AI generated art community is one being pushed by corporations while all the 'lowly'/proper artists want nothing to do with it. Really makes me shake my head seeing the all the indie collaborative efforts of this community going back years, wrangling programs and notebooks for the sole purpose of putting out out-of-the-world artpieces.

[@LarsRichterMedia]

Here is my issue with OpenAI: Their white paper says they collected about 400 million images from the internet (doesn't specify anything further than that) to train DALL-E 2. I've already seen the first article by MIXED calling it "OpenAI's first money printer" because they plan to sell access to it. What do the hundreds of thousands of creatives receive who produced these hundreds of millions of images? Nothing. They weren't even asked for permission. You can claim they uploaded their work to a public space, so, hard luck! I would call that an utterly naive understanding of what copyright law should mean in the age of AI if it is supposed to mean anything. Without the gigantic amount of time, effort and money all these human beings put into their craft (photography, painting, designing) to produce all of this data, Dall-E 2 or even Dall-E wouldn't exist at all. Someone explain to me how it is morally OK for OpenAI to be the sole profiteer here?

[@edd542]

@@LarsRichterMedia Well humans do essentially the same thing as these AI's. The art generated by artists is influenced by the artists they look up to, their life experiences, their visual library (which includes memory of photos viewed on certain subjects). We too are a collection of what we have seen including other peoples work.

[@LarsRichterMedia]

@@edd542 declare these AI systems to be conscious and to have a right to perceive the virtual public space as a sentient being and there's (almost) no issue. I'm an artist myself and I have spent years coding. I know both worlds well. Well enough to form my opinion that these AI systems like Dall-E 2 should not be considered as doing what humans do when they're inspired by other artists. No, it's a data harvester feeding a process that creates a latent space from which original visuals can be drawn. It's a machine that has no consciousness and is built on the unpaid labor of countless artists, photographers, designers. Especially if a legal precedent is set that says humans and AI are legally doing the same here it would be an absolute disaster.

[@edd542]

@@LarsRichterMedia it doesn't matter if they're conscious or not. I am also an artist and I can tell you that me and a lot of other artists as well rely on reference images for our design, the AI essentially does the same thing. You could also say we also benefit from the unpaid labor of photographers we use for reference...

[@LarsRichterMedia]

@@edd542 ok, my opinion is different than yours. To me it does absolutely matter. Just because no individual lives in a vacuum doesn't mean we need to simply accept such AI systems to bulldoze parts of our economy with their ridiculous power. They are not doing the same because they're nothing more than sophisticated math equations that can process data hyper fast. An artist selecting a few references is not completely copying all the data from insta, artstation etc. into a machine that then spits out originals. I hear you saying humans are also nothing but machines just worse ones in this and many other aspects? Here is where I draw the line for now. Unless you call these systems conscious beings they don't have a right to harvest public spaces without consent. Especially because no artist in the world uploaded their work and filled such public spaces with billions of images thinking that in a few years some AI will just scan it all at once. Equating AI to humans in this case to me is an absolute disaster. AI is not "stealing like an artist" because it is not an artist, it's not a conscious being that gets inspired by other works. It's a machine that was literally built to a huge extent by unpaid creative people. I don't know what artist you are but the small bits and pieces I take as inspiration from others more often than not also lead to me being transparent about who I've been looking up to but even if I wasn't the comparison between human inspiration and AI harvesting and processing I find extremely misleading.

[@3rdstrikestan930]

there goes my future career

[@ghostnights8538]

Glitchy