Тёмный
No video :(

ControlNet Revolutionized How We Use AI To Generate Images 

bycloud
Подписаться 152 тыс.
Просмотров 107 тыс.
50% 1

Check out Opencv's New AI Art Course Kickstarter at bit.ly/410U3Xs and join their AI Art Contest now to win an iPad Air! form.jotform.c...
ControlNet: Adding Conditional Control to Text-to-Image Diffusion Models
[Paper] github.com/lll...
[Code] github.com/lll...
Style2Paints v5 Preview
[GitHub] github.com/lll...
This video is supported by the kind Patrons & RU-vid Members:
🙏Andrew Lescelius, Chris LeDoux, Shawn77134, Panther Modern, Jake Disco, Demilson Quintao, Tony Jimenez, Shuhong Chen, Hongbo Men, happi nyuu nyaa, Carol Lo, Mose Sakashita, Miguel, Bandera, Gennaro Schiano, gunwoo, Ravid Freedman, Mert Seftali, Mrityunjay, Richárd Nagyfi, Timo Steiner, Henrik G Sundt, projectAnthony
[Discord] / discord
[Twitter] / bycloudai
[Patreon] / bycloud
[PayPal] paypal.me/bycl...
[Profile & Banner Art] / pygm7
[Thumbnail Result] Taken from Toyxyz on Twitter

Опубликовано:

 

13 фев 2023

Поделиться:

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист
Посмотреть позже
Комментарии : 326   
@bycloudAI
@bycloudAI Год назад
The artist community is probably going to go insane about this line art colorization tool now, hopefully it'll be positive D: ad: Check out Opencv's New AI Art Course Kickstarter at bit.ly/410U3Xs and join their AI Art Contest now to win an iPad Air! form.jotform.com/230370613300439 And check out my Patreon if you wanna support my work!! www.patreon.com/bycloud
@fresh218
@fresh218 Год назад
Surely they won't lose their marbles, *surely*
@imsonoided
@imsonoided Год назад
I'll die before I ever use AI in any work I do. AI is a crutch for people with no passion, talent, or new ideas.
@user-yd4om1qw3n
@user-yd4om1qw3n Год назад
@@imsonoided ok boomer
@Amelia_PC
@Amelia_PC Год назад
As a comic book artist, I can say it'll make my life way easier. Usually, my gigs are interior art (line art) and colors. If this ControlNet with Anime Line Drawing can preserve the lines and color the thing, I'll be more than happy with it!
@peterbelanger4094
@peterbelanger4094 Год назад
Stable Diffusion is quickly becoming the 'Blender' of ML image creation.
@JCTechCorner
@JCTechCorner Год назад
Time to give ControlNet my childhood notebook and see how my doodles look with its artistic touch
@westingtyler1
@westingtyler1 Год назад
do it
@wumpusrat
@wumpusrat Год назад
I had an old graphic novel I was working on ages ago, where I'd just laid out all the sketches of the scenes and how I wanted the characters moving, but never actually did any fine work myself (though I'm not that great an artist). I'd be really curious to see how something like this would generate my scenes.
@westingtyler1
@westingtyler1 Год назад
@@wumpusrat totally. then you could easily publish it on amazon.
@arakemi1080
@arakemi1080 Год назад
ControlNet ain't an artist yet, don't call it as such.
@Osjey
@Osjey Год назад
as an artist myself, the Lineart to Color would be huge. Color and Rendering is mostly fundamentals and technique and is not the most important part of anime artwork. Even though the drawing is the most imporant part of an image, coloring it takes just as much time. So being able to only focus on the drawing itself would be such a relief.
@glint6070
@glint6070 Год назад
My hope is that this increases interest in making line-art AI; it seems relatively obvious that outputting high-quality line art, then colorizing that line-art would allow for a very easy way to maintain coherency By extension, high-quality text-to-2D animation can finally be realized
@averesenso
@averesenso Год назад
If you like doing line art publish line art
@imsonoided
@imsonoided Год назад
Or, get this, you are too lazy to learn color theory and don't actually care about making art.
@Cyberlong
@Cyberlong Год назад
@@averesenso bruh?
@YuriRosas_
@YuriRosas_ Год назад
Really. Imagine new seasons of Anime being released in a shorter time.
@Valensiakol
@Valensiakol Год назад
As someone who can do lineart but sucks at coloring and shading, I have great interest in this tech. It's something I've dreamed about for ages.
@HCforLife1
@HCforLife1 Год назад
So true - I am having ton of sketchbooks waiting to go throught this lol
@peterbelanger4094
@peterbelanger4094 Год назад
I love coloring and shading, and I love using this extension. It's great for everyone! had no trouble installing it too, runs fine on my 1060GTX w/6GB vram. (faster without the 'lowvram' option)
@Valensiakol
@Valensiakol Год назад
@Kozzie Tea I watch tons of videos like that. Of course I will still continue to develop my own skills, but that doesn't mean I can't appreciate new art techology like this.
@wumpusrat
@wumpusrat Год назад
Same. I can do basic line art, but I suck at proper shading. It would be really fun to see how an ai would "finish up" my sketches and line art.
@cybermollusk
@cybermollusk Год назад
I like that these AIs are slowly starting to cater to actual artists instead of just people with no art skill. I'm an artist and I want to use these in my work, but I need far more control than just typing words in a box. I want a photoshop-level of control along with these AI generative abilities. I'm glad to see this is slowly starting.
@xn4pl
@xn4pl Год назад
It wasn't ever just words in a box, there's img2img and as an artist it should be way easier to use, make a rough sketch, feed it into img2img for adding details and select parts you like from multiple generations and composite them in your image redactor. You can work on some nuances while it generates, like fixing hands or paint over artifacts to fend off picky reddit mods.
@grayghost7926
@grayghost7926 Год назад
It's been really hard getting people to realize it wasn't just a few words to generate good images lol. Actual artists had to wade through comments like yours as they dealt with threats and hate helping test and helping design and develop these types of tools in a way that artists could use. lol... I mean... they always did cater to artists and were being trained by artists and used by artists and innovated by artists... how do you think these tools keep coming about? The artists have been making requests and pitching ideas. While non artists... well pretty much spewed hate everywhere to the point the word "Artists" has become derogatory in a lot of circles lol.
@cybermollusk
@cybermollusk Год назад
@@grayghost7926 😳 why are you so offended by my comment? I don't think there was anything offensive in it. You said "Actual artists had to wade through comments like yours as they dealt with threats and hate." I didn't say anything hateful, threatening, or negative about artists or anyone else. I was saying I'm happy that they're catering to artists more. Most of these AIs only had txt2img at the beginning. They added img2img later. So no they were not catering to artists. They were catering to the general public. And that's fine, but an illustrator doesn't want to trade drawing for typing. But now they are creating more tools for us artists and I'm happy about that.
@cybermollusk
@cybermollusk Год назад
@@xn4pl Almost all these powerful art AIs (SD, Midjourney, Dall-E, etc) started as only txt2img. So yes, originally it was only typing words in a box. No illustrator wants to just type. Img2img is a more recent development and it makes these AIs more useful to experienced artists. Img2img is part of what I mean by it catering to artists now.
@RoddKane
@RoddKane Год назад
@@grayghost7926 Discovery isn't art
@alankun1716
@alankun1716 Год назад
Despite my initial concerns with the speedy developments of AI Art Generation and my initial persistent avoidance of anything that has to do with AI. I couldn't help but be curious to see what all the fuss is about and decided to finally get my hands on one, downloaded a couple pre-trained models I found online, until I eventually found myself in another rabbit hole. I've realized how great these were with helping me visualize any character design ideas I'd have before I proceed to sketch out these ideas I've accumulated. I basically treat these generations like I would of a normal photo reference, but instead of having to possibly dig for hours on the internet for a photo that I would need, I could save a lot of time with these as I keep them running in the background while I do other related work. I've came across other similar tools like the ones shown on the video, but they weren't that great so I couldn't find any use for them. I frequently make line sketches of my ideas before I throw in a bunch of color mess in attempts to plan out which colors I'd use on the final output. I was thinking if I could manually choose the color scheme and theme It would use to fill out the lineart, it would be tremendously helpful for me. It's really hard juggling my time between my university work and my illustration hobby, I'd appreciate anything that would speed up my workflow.
@shubashuba9209
@shubashuba9209 Год назад
Say goodbye to the people who thought they could have a job being "prompt engineers". The real money makers are the products that allow artists to refine their work down to a granular level with perfect control, not the people who type in random words with random results and try to piecemeal stumble their way into a good prompt.
@Thesamurai1999
@Thesamurai1999 Год назад
Exactly.
@nickchua5772
@nickchua5772 Год назад
Artists can also feed their own best works into the machine to replicate their style and completely avoid some ethical concerns
@Natsukashii-Records
@Natsukashii-Records Год назад
@@nickchua5772 Nope, you need to use an already made model to train on top of it. A model that has been trained in art 'unethically'. Regardless most people don't give a fuck. :^)
@BumboLooks
@BumboLooks Год назад
It is literally nothing more than a tool to illegally plagiarize art with extreme lossy compression in multi gig "models".
@hangduong9608
@hangduong9608 Год назад
I think an artist who has AI knowledge is better than one who doesn't
@polarzup6034
@polarzup6034 Год назад
This basically enables you to get a reference from where you want your image to end, allowing you if you want, to understand the color relantionships that are set and learn from it while adding the colors by hand Infinite study template babyyyy
@dee6561
@dee6561 Год назад
This is simultaneously painful and amazing as an artist. I can't tell whether I should have this in my workflow, or demonize it all together. I will likely be forced to do the former to keep up...
@zenu903
@zenu903 Год назад
Imagine coloring mangas with this
@GoharioFTW
@GoharioFTW Год назад
H
@cybermollusk
@cybermollusk Год назад
Imagine every frame of an anime looking like the high quality cover art.
@cybermollusk
@cybermollusk Год назад
@@diowoh I know. It's not quite there yet. I've seen animations done with AIs. I'm guessing you have too. You get flickering and noise. But I'm sure they'll figure it out sooner or later.
@strafeeto
@strafeeto Год назад
@@cybermollusk It would have to color 5000-30000* or more frames per episode, it probably wouldn't be realistic. Everything would have to be uniform and that's not even mentioning the backgrounds.
@arnowisp6244
@arnowisp6244 Год назад
@@cybermollusk Corridor Crew figured it out already. They made a video using Stable diffusion to make there anime using their live action as reference. Apparently they use a an antiflickering Plugin to fix it.
@AparnaModou
@AparnaModou Год назад
We now have more tools to use. Not just generating images on image generators such as Bluewillow, we still have our own identity by using our skills to edit or just simply using a these images as inspiration.
@jofreproblem
@jofreproblem Год назад
Stable Diffusion hits another win. I hope OpenAI learns from this.
@4.0.4
@4.0.4 Год назад
I don't think they will; DALL-E 2 hasn't had any improvement since the beginning.
@peterbelanger4094
@peterbelanger4094 Год назад
Autodesk never learned from Blender.
@jofreproblem
@jofreproblem Год назад
Watching out what the CEO of OpenAI says about when it will come out the source code of ChatGPT, it's kinda hopeless.
@Cubicflow
@Cubicflow Год назад
@@peterbelanger4094 An comparison i didnt expect but that i totally agree with, autodesk sucks ass lol
@arnowisp6244
@arnowisp6244 Год назад
@@peterbelanger4094 Problem us that Autodesk is industry stand so your Blender skills are likely non transferable in the Industry sadly.
@neutronine
@neutronine Год назад
Always informative and entertaining. Thank you
@swannschilling474
@swannschilling474 Год назад
Peeps its already up as an addon in 1111!! 🤯
@liminalai
@liminalai Год назад
The editing on this cracked me up🤣 Great video👍🏼
@SylvesterLazarus
@SylvesterLazarus Год назад
5:25 Kudos to Illyasviel for not releasing the coloring tool until ethical issues are resolved!
@SylvesterLazarus
@SylvesterLazarus Год назад
@Singularity There were many people in history who said that to many things.
@SylvesterLazarus
@SylvesterLazarus Год назад
@Singularity Who only knows the hammer will see everything as a nail. So what do people know who call everything virtue signalling? Lies and desception?
@YVZSTUDIOS
@YVZSTUDIOS Год назад
ngl, for me as an artist, this sounds like one step closer to where I just have to draw the lineart of an animated (anime style) sequence and let the AI color and shade it. But yea, that will require temporal coherency and so on. Well, ofc the ultimate tool of the future would be what I just explained plus: me guiding the AI what colors to use with a single reference sheet and then the possibility to specify light source(s) in 2D/3D space, and the AI shades it accordingly. THAT would be absolutely amazing. And I'm not sure how the ControlNet works with Stable Diffusion. Does it require a custom base .ckpt? Because it would be absolutely amazing if I could use my own custom trained models, bc I've trained them on my own artstyle.
@imsonoided
@imsonoided Год назад
Why even bother making the line art? Just push a button and you could get the line art too, right? You'll get to call yourself an artist and not actually pick up a pencil or even move your mouse.
@YVZSTUDIOS
@YVZSTUDIOS Год назад
@@imsonoided See, that's the interesting part. One of my friends who has many years of experience in the creative industry, said this: In the near future regular artists will be able to do what a creative director does with the help of AI tools, creating works of art that are designed by us artists. The barrier of entrance to create comprehensive artworks will drop. If this AI development doesn't slow down, it really will be like what u say. Everyone will be able to make their of animated show or comic with one push of a button or some mind reading tech. But that's just the craftsmanship part of art. The artistic intent will still come from us human artists. The definition of what is art is currently changing. But it will take some time to define rules for ethical and legal usage of AI art. But think about it. This very same thing happened when 3D/CGI was new. Many people called it cheating, no real art, etc. I mean some still say that to today's artists doing digital paintings etc. History repeats itself everytime a new technology appears. It all comes down to why do I make art? To tell a story? Share my story? Self expression? That's different for every artist.
@bobsmithy3103
@bobsmithy3103 Год назад
@@imsonoided Because it allows you to make exactly what you want, words are a lossy medium. Like sometimes you just want that control to exactly create your vision. Like sure, for a consumer of art, the end result of art done by hand vs art done by ai may seem the same, but I wouldn't say the same is true for the artist. However for an artist to exactly create their vision they'll probably have to toil for half a decade or more years to reach that level. And unfortunately, with the advent of ai I can see a lot less people reaching that level simply because pursing art is an even worse investment now.
@idk_what_im_doing04
@idk_what_im_doing04 Год назад
@@imsonoided whats the point of that, though? Theres no real effort or inner emotion to express unless you make your hand dirty.
@Dooshanche
@Dooshanche Год назад
@@YVZSTUDIOS it is as if we will be able to make all types of interactive media in one place and share it with each other, now if we could only come up with a name for such a place...hmmm...
@Agony.
@Agony. Год назад
Bruh, people think they are talented just by typing prompts 🤦‍♂️.
@frostreaper1607
@frostreaper1607 Год назад
I don't draw classic manga but if this V5 Dorothy model ever gets released I want to test it on my own line art and start drawing more manga like, this might convince me to make comics, as those things are incredibly labor intensive. Don't have high hopes for unique species and fantastical beasts tho.
@Descalibrado_
@Descalibrado_ Год назад
YESS IVE BEEN WAITING FOR ONE LIKE THIS SINCE STYLE2PAINTS
@NotASpyReally
@NotASpyReally Год назад
I don't mind AI art, it's inevitable, we have to accept technology. What I don't like is that people are going to lie (and probably already lying) about AI art being Human art. I'm gonna wanna support human artists and people are gonna lie.
@highspeedrailenjoyer1045
@highspeedrailenjoyer1045 Год назад
I believe in Lvmin Zhang supremecy.
@EverythingBattletech
@EverythingBattletech Год назад
This would be great for our Battletech Line Art. Can't wait to be able to use it
@StrangerHappened
@StrangerHappened Год назад
This cute plushy Cloud surrounds us with its informative fluff.
@stanleywtang
@stanleywtang Год назад
With over 20 years of experience as a visual artist and animator, I've only used text2img to play around with ideas, as using text doesn't feel natural to me. Hopefully ControlNet can change that
@dystopiaproject
@dystopiaproject Год назад
I didn’t understand good how it works, but it seems you still need a text as a prompt even with the line art ? No?
@ByronDZero
@ByronDZero Год назад
I would absolutely love to have something like this to help with the production of my web comic
@arnowisp6244
@arnowisp6244 Год назад
I suggest not to mention to anyone you used AI else you'll be branded for Heresy.
@zanly5039
@zanly5039 Год назад
​@@arnowisp6244 nah fuck that, imma let everyone know how AI art can empower individuals to make things that used to require entire studios. more and more ambitious passion projects are becoming possible.
@Solizeus
@Solizeus Год назад
Being able to train it to draw a character models and objects in different positions and then being able to pose it in any position you want it to draw may make animation studios do really well, or at least webtoons
@BumboLooks
@BumboLooks Год назад
It doesn't work very well.
@pinpointping6175
@pinpointping6175 Год назад
Dude. Next, 3D scene to image; rendering in seconds!
@Thesamurai1999
@Thesamurai1999 Год назад
How about images to 3d scenes?
@Eren_Yeager_is_the_GOAT
@Eren_Yeager_is_the_GOAT Год назад
@@Thesamurai1999 that already exist
@kw4093-v3p
@kw4093-v3p Год назад
Love OpenCV! One of the sponsored I can get behind.
@FaizalKuntz
@FaizalKuntz Год назад
i want my manga to be color but consistency in coloring manga is a huge problem so if this AI could help colorize (at least coloring base color) it would take off heavy burden artist have and since the color already applied the artist only need to focus to tone the color and highlight the important part.
@1dgram
@1dgram Год назад
What were the ethical concerns?
@BlueSatoshi
@BlueSatoshi Год назад
Tracing other people's work, probably.
@1dgram
@1dgram Год назад
@@BlueSatoshi You mean tracing via edge-detection followed by using the AI to plagiarize the artwork?
@BlueSatoshi
@BlueSatoshi Год назад
@@1dgram I dunno the specific methods that would be involved, but essentially tracing and otherwise straight up plagiarizing other people's work. Something well beyond simply generating a random image with a model that was trained off a million or so different pieces of artwork.
@cesar4729
@cesar4729 Год назад
Honestly, the canny control already do that (kinda), so im a little confused too.
@klaushermann6760
@klaushermann6760 Год назад
I don't read manga because it is always in black and white. But now AI can make them colorful for me.
@arnowisp6244
@arnowisp6244 Год назад
AI making Manga Colorful has been a thing for a while now before Dtable Diffusion.
@devinmausia4939
@devinmausia4939 Год назад
I can't wait till i can speak it and then see it in augmented reality!
@jaceg810
@jaceg810 Год назад
3:34 still does not know how many fingers a human has, we are still safe for now
@twitterglobalarmy
@twitterglobalarmy Год назад
With all these new technologies, I already feel behind having only begun to understand AI's possibilities. It would be wonderful to see another free video about Blue Willow in the future.
@BumboLooks
@BumboLooks Год назад
I've tried it all recently. You aren't missing out on anything.
@MrEdukada
@MrEdukada Год назад
I love that AI hands are getting normal
@thegeekclub8810
@thegeekclub8810 Год назад
I think is has the potential to be quite helpful, and I wouldn’t blame any artist who uses this, but I’m still uncomfortable with using a system that uses art from artists who didn’t consent to having their art used in training. We can argue til the cows come home about whether or not the AI is operating in a way similar enough to a human to qualify as having been “inspired” by other’s artwork, but I’m still uncomfortable with the way this technology is going to take artists’ jobs off the back of their work. We should be facilitating human creativity, not outsourcing the creativity to a machine.
@cesar4729
@cesar4729 Год назад
Creativity is inherent in humans, not in a group of "artists". Believing that bringing more people ways to experience their ideas is reducing creativity is as arrogant as it is discriminatory.
@ShawnFumo
@ShawnFumo Год назад
Thanks for trying to be fair in your opinion. I don't think there is easy answers about any of this, and there is a lot of be uncomfortable about. On the good side, I've definitely seen people getting in touch with creativity and artistic sensibility through AI ironically enough. Speaking just for myself, I didn't put a lot of time into drawing/painting partly because I didn't even know what I would want to make. Experimenting with AI, I quickly got bored with making random funny things or generic images that you see constantly. So I kept experimenting and curating the results and found things that personally interested me to be brought into the world, vs just images from other people I like to see. I actually feel more inspired now to go more traditional with art skills, to get that additional involvement and control. This can also be quite helpful for things like art therapy. A lot of people use art to get through difficult emotions and experiences and you can have some of those benefits even using AI, seeing images appear that start to express what you're feeling. On the bad side, even putting aside artists' work used in training, job security, etc there is various other issues. If you go into any space that lets you browse the profiles of people who are using the system and it is a space that allows adult content, you'll pretty quickly realize that anyone can create horrible stuff like cp by just typing a couple of words, using the normal models everyone uses. And Dreambooth and Lora allow you to train someone's face easily, with just a few photos. As the photo-realism of the models increases, the idea of "deep fake" as a separate concept starts to break down and anyone can show anyone doing anything else they want, at the drop of a hat. This kind of power has the potential to be very corrosive IMO.
@BrgArt
@BrgArt Год назад
@@cesar4729 Childrens are extremly creative. even more so than adults in fact. why are childrens not leading the entire art industry ? Because it's not controlled and doesn't mean anything at all. It's not refined. Refining your imagination takes years of hard work and patience. Something only artist, architect and engineer have.
@nikki-zeero3122
@nikki-zeero3122 Год назад
@@cesar4729 you are right about the "humans" part. Wrong about the rest . AI isnt an artist or a human . Its an automation bot , that just guesses the best output depending on the input without having any awareness on what even the input means . So no , doesnt deserve the same rights for "creativity" that us humans have .
@cesar4729
@cesar4729 Год назад
@@nikki-zeero3122 It would be interesting to know what the "wrong" part is, because nothing you said has anything to do with what I said. Have you even read?
@pastuh
@pastuh Год назад
Already using, insane how good.
@pxrposewithnopurpose5801
@pxrposewithnopurpose5801 Год назад
colored uncencored doujins are all i need
@maxmustermann3938
@maxmustermann3938 Год назад
Messed around with this a bit and it is very promising, but I am really missing that color control. Specifying colors in the prompt is super unreliable - if you specify clothes, hair, eye color and so on it'll just mix things up. Skin color seems to be extremely unreliable too. So a sketch+color thing would be amazing this.
@strawberyberries
@strawberyberries Год назад
This sounds amazing!! Now ill finally have refrences when i color
@AubreyWatt
@AubreyWatt Год назад
Could someone tell me if you can run stable diffusion from external hard drive? All these addons and models are starting to take up a lot of space lol
@tellu5493
@tellu5493 Год назад
Yeah it'll run there just as well as anybody else.
@AubreyWatt
@AubreyWatt Год назад
@@tellu5493 can I just drag the stable diffusion folder into my external or do I need to do any calibrations to the program? Or anything extra besides just moving it?
@DJVARAO
@DJVARAO Год назад
Yep
@Slav4o911
@Slav4o911 Год назад
@@AubreyWatt You can't just copy the folder, you have to Install it in a folder on the external drive. Then you can copy or move only the models from your old folder.
@user-bv5bz2kz4t
@user-bv5bz2kz4t Год назад
you can use symlinks
@arnowisp6244
@arnowisp6244 Год назад
Control Net debunks the Notion that AI artist are not creative because you are commissioning the AI the make the Art for you. Not you making the art. Control Net allows you to co create the art with the AI now.
@APrettyGoodChannel
@APrettyGoodChannel Год назад
I don't think a 512x512 depth map would have any advantage over a 64x64 depth map, because internally Stable Diffusion works at 1/8th resolution using a sort of compression system, and so any 512x512 input is turned into 64x64 chunks anyway.
@lennytheintern
@lennytheintern Год назад
Bro, I swear. this feild moves too god damn fast.
@smartduck904
@smartduck904 Год назад
But can it not color over frames and time
@grindermachine8856
@grindermachine8856 Год назад
That would really be helpful to comic/webtoon/manga artist to more efficient
@arnowisp6244
@arnowisp6244 Год назад
Webtoons have generic Colorong anyways.
@XonTheOne
@XonTheOne Год назад
im creating a very long story, do you know if its ok to color pages, refine them in Photoshop then post the 120 chapters with the intent to make money ?
@SavoryParty
@SavoryParty Год назад
I'd recommend looking for videos explaining copyright ai stuff if that's what you're talking about
@SENYSENofficial
@SENYSENofficial Год назад
know this is a game change ! it is a real tool for artists and animators
@codelessunlimited7701
@codelessunlimited7701 Год назад
Gemini main structure is focused on digital neural networks.
@paintangel13
@paintangel13 Год назад
The way I see it, they are just getting dumber and dumber. And these AI programmes are helping us to lose more and more weight, which means we are becoming more and more unlearned. They take away our ability to find our own process for problems and thus to become better and better. At some point, no one will learn colours and light, because the AI will take that away from them. But since it can't really think, everything will look generic. Like it does now.
@arnowisp6244
@arnowisp6244 Год назад
That's why artist are still needed. To get the parts AI cant figure out. Don't be a luddite and accept Progress.
@nikki-zeero3122
@nikki-zeero3122 Год назад
@@arnowisp6244 yall dont even KNOW what the term luddite means and its funny really . In history , the "luddite" people werent mad because of "oh no technological advances!!!'' They were mad because of what people and their bosses did with those advances : fired them without a second thought, saw them as objects to be replaced , and underpayed+mistreated the still employed ones horribly . Much like what you guys are doing really - half of the "ai bros" advocate : You "artists" are done for! You have no power and you are nothing agaisnt this new fart generic thing we have now ! Your skills dont matter and every single one of you will die unemployed! ( and some going as far as to tell artists to k1ll themselves for it, haha .) Please do your research before deciding on a view ,as you tell artists' to do when they spill some logic to your ears about this ethical issue. Because it seems like you guys have already hit the "the ai will make us forget how to learn and think" step . Its embarrasing for everyone.
@Veptis
@Veptis 8 месяцев назад
It's all done by a aibgle student? I hope this gives them many opportunities down the line.
@draken5379
@draken5379 Год назад
You dont need to train anything, it works with any SD checkpoints.
@alchemyarthouse
@alchemyarthouse Год назад
Any workarounds for getting ControlNet and StableDiffusion on a Mac?
@DM-dy6vn
@DM-dy6vn Год назад
Hello AI Overlord!😊
@gametram6139
@gametram6139 Год назад
this is the only time where i can actually say ai art is bad. Image generation is meant for visualization, while this is just laziness. Im not at all the kind of person to say "ai art bad its gonna replace all of us and this is a detroit become human scenario!!!!!!!!" and i actually think its amazing, but only when its used as a tool, not this. I dont blame you though, you need some kind of content to make on an ai related channel. You have gained 1 dislike and 1 subscriber.
@CharmingGal
@CharmingGal Год назад
This will definitely help me to improve my drawing skills ❤️
@BumboLooks
@BumboLooks Год назад
You will never be able to draw...
@feardietester5507
@feardietester5507 Год назад
You will never be able to draw... (2)
@CharmingGal
@CharmingGal Год назад
@@feardietester5507 okay believe,what you want to believe. But that never stop me because I have faith within myself, have a loving day 💝💘 ✨ I forgive you both of you✨
@lilianchin2773
@lilianchin2773 Год назад
Is it possible for the ai to colour the line art with only flat colours?
@heretolearn4214
@heretolearn4214 Год назад
Don’t know what I just heard. Guess this is what feeling stupid is like 😢
@AsU-yz9lo
@AsU-yz9lo Год назад
Bro me too
@Thesamurai1999
@Thesamurai1999 Год назад
Lack of knowledge does not equate to stupidity. You simply gotta learn, read more and familiarize yourself with the subject.
@Wolgan
@Wolgan Год назад
Thanks man!
@gregorvm7443
@gregorvm7443 Год назад
I'm really surprised that still animation studios haven't use this to make the animation process more quickly, this technique as it is, won't work of course, but if you add somehow temporal coherence to the colorization process, you can fire all people that only colorize the anime stuff thus making the process faster thus making more money, faster. And if for some reason it doesn't work perfectly you can use it to colorize in between frames I think it should work fine there with temporal coherence because the IA could use the previous frame to make the colorization more precisely.
@nocturne6320
@nocturne6320 Год назад
Temporal coherence is a giant roadblock and researchers have been trying to tackle it for years, that still won't be solved for some time
@gregorvm7443
@gregorvm7443 Год назад
​@@nocturne6320 Yea I agree, even audio and video models nowadays have a very hard time with that, but still, I think if the frames are very similar, and you just pass the previous frame plus the exact draw of the next frame with its shadows and so on, it's going to be possible to get the next frame colorized more precisely though it's not easy at least I think it cannot be done without re-training a completely different model just for that.
@imsonoided
@imsonoided Год назад
Ever thought that animators actually enjoy the process and don't want to automate the fuck out of it?
@MrFram
@MrFram Год назад
@@imsonoided > people enjoy in-betweening t. room temperature in celcius iq
@dibbidydoo4318
@dibbidydoo4318 Год назад
@@imsonoided not in japan where they're overworked for little pay.
@zyxwvutsrqponmlkh
@zyxwvutsrqponmlkh Год назад
>How many waifus do you need? There can never be enough.
@arnowisp6244
@arnowisp6244 Год назад
I want to draw more Husbandos. As A Guy, I like women but too much and it feels like the Winx club.
@derekborders9647
@derekborders9647 Год назад
Isn’t that locked/trainable setup what “transformer” means?
@kukukachu
@kukukachu Год назад
So people are gonna bitch about their work being used, but then use A.I. to color in their drawings?....the world needs to grow up.
@nikki-zeero3122
@nikki-zeero3122 Год назад
Nah really . The world needs more ethically sourced AIs instead of stealing around the whole world's worth of art and turning them into generic automations that people mistake for art due to the way it is advertised . If any AI is sourced by artists' legit and legal consent , knowing what it will be used for - I am more than happy to see it being used by whoever the fuck ! Could even use it myself to GUIDE myself on ACTUALLY drawing and painting . People cant have copyright over the generated pngs that AI makes after all , so it would be super cool to see actual artists use them to learn rather than sad pathethic waste of people that call themselves "prompt writers" that sell those pics by marketing them as "human art" and keep throwing around the word 'luddites" without even properly knowing what it means ,lololol .
@kukukachu
@kukukachu Год назад
@@nikki-zeero3122 Disagree. There's not any stealing happening. A person, looks at art and then uses that art as reference to theirs...is that stealing? No. The A.I. uses the art uploaded to it as reference and creates its own thing using those references. Even if the picture looks like a blatant copy, it's not.....humans do this as well. A.I. and humans are not too far off from each other in how they function. The only difference is that humans have emotions that go with their programming. If you look at what the A.I. is doing compared to what humans do, you would see that there is some DEEP hypocrisy happening in this argument.
@nikki-zeero3122
@nikki-zeero3122 Год назад
@@kukukachu also please stop using cheap words such as "hypocrites" and "luddites" or whatever . The real hypocrisy is the ai bros guys calling themselves an artist without even being willing to put blood and sweat into it . And the luddites- you guys dont even know what that word actually means . You look like a child that has learned a swear word and is sticking it to everything it sees -mature up please . And the real meanning of the word is also explained in the video, so i highly advise you and many others look at it .
@nikki-zeero3122
@nikki-zeero3122 Год назад
@@kukukachu I linked it , but it appears youtube didnt let the comment go through due to the link. The video is named "AI art will have consequences" . Please look it up . It really clearly states legit arguments and facts at everything you guys have been blinded about by the advertisement of these png generators .
@kukukachu
@kukukachu Год назад
@@nikki-zeero3122 I didn't ever use the word "luddites".
@IsaacFoster..
@IsaacFoster.. Год назад
As an average weeb, I'm absolutely hyped about the new AI art projects. Although I don't have enough resources for making AI art myself, entering Pixiv with the AI tag feels like diving into heaven.
@4.0.4
@4.0.4 Год назад
You can use Colab, if you don't have a good enough graphics card.
@Lincedehielo
@Lincedehielo Год назад
Can you imagine, export the frames of, let's say, Kamui, Kimba or the first Sanctuary saga of Saint Seiya, run gimp batch scripts to get the line art and then re-colorize them to a Katsuihiro Otomo/Masamune Shirow style? Keeping the original line art style *¬*
@joeshmoe4207
@joeshmoe4207 Год назад
Check out stable horde
@kimjohn3877
@kimjohn3877 Год назад
Can i make my own manga using this? I have amd gpu bdw
@FieryRMS
@FieryRMS Год назад
Hey the OpenCV contest already ended 2 days ago and the winner probably got announced as well💀
@Johnnyupside
@Johnnyupside Год назад
Oh hey I've seen some peps in the forum talk about this before. Yeah it's pretty niffty
@hdtv4ktv408
@hdtv4ktv408 Год назад
What App Can Colorized Photos On Driod?
@Yue4me
@Yue4me Год назад
where is the thinking part
@unknownrocketeer9289
@unknownrocketeer9289 Год назад
love your content! thanks for keeping the boomer mid 20s y. o. net programmers in the know lmfaoooo
@holimount9311
@holimount9311 Год назад
The only concievable way AI "art" ever gonna have any kind of future is by creating something anything that gives any value to real artists instead of just vomiting cheap copies of their work.
@Noah-ws8ho
@Noah-ws8ho Год назад
Did you... watch the video at all?
@BrgArt
@BrgArt Год назад
@@Noah-ws8ho yes and as an artist i still see an automated process that give you a final result instead of a suggestion. It's already hard enough backtracking my own work and changing things around once i settled for something. If this keep on going we wont ever see any innovative idea for centuries to come. art is a process.
@yesman1743
@yesman1743 Год назад
This could help me to make colored high quality manga or comics with a God speed level.
@Trivia2023
@Trivia2023 Год назад
Nice video
@onlyyoucanstopevil9024
@onlyyoucanstopevil9024 Год назад
REMEMBER A.I HATER, THIS IS JUST BEGINNING REVOLUTION HAS BEGUN !!!
@Ndik
@Ndik Год назад
Now I can create a kawaii waifu without having good drawing skills 🤣
@sourcerror
@sourcerror Год назад
You can already do that with NovelAI, Anythingv3 and WaifuDiffusion.
@maestro3676
@maestro3676 Год назад
@@sourcerror There is Anything v4.5, and many other anime models that are far better than waifu diffusion, like OrangeMix models.
@davidscann
@davidscann Год назад
@@maestro3676 p.s anything v4.5 was a troll mix by andite consisting of anything v3.0 and aom2 it produces OKAY results but it doesn't take away from the fact that andite didn't put any care into it which is funny but yknow (it wasn't supposed to be good)
@taihuynhuc3135
@taihuynhuc3135 Год назад
@@davidscann Actually a mf from UD told him about a very good mix of AOM2 and his finetuned model. He tested it and found out that is true. So he decided to make some small changes to the recipe and then called the new model as AnythingV4. Even the man who uploaded Anything V3 to HuggingFace and make it popular on there encouraged him. Tbh, Anything V4 looks good and I don’t really mind about whatever it is a troll model or not. And Anything V3 is a mix anyways
@davidscann
@davidscann Год назад
@@taihuynhuc3135 I mean, my source is andite himself, so do with that as you will
@RemitheDreamfox
@RemitheDreamfox Год назад
i always wanted a cloud to talk tech to me \uwu/
@kridlob5264
@kridlob5264 Год назад
Like WOW Open source? Capable of finishing my drawing? Running it on potato pc? (maybe) Sign me in NOW
@XxXVideoVeiwerXxX
@XxXVideoVeiwerXxX Год назад
Ah style2paint I thought it stopped getting updated
@gomezabel8880
@gomezabel8880 Год назад
How can I buy a stake in this company?
@Noah-ws8ho
@Noah-ws8ho Год назад
It's the result of university research, not a company. So as of now just freely available, nothing you can invest into, would have to ask the researchers that made the improvements if they want to start a company lmao.
@atoma_5176
@atoma_5176 Год назад
I just yesterday installed stable diffusion on my desktop and I have no idea what it does, what are it's alternatives or possibilities, and I don't know if I should get into it or not, as I just use it to generate hentai with anythingv3. I feel like I'm missing out on something grand and epic but I don't know where to start or if it's a right decision, weirdly enough it gives me anxiety knowing this super large community exists
@ShawnFumo
@ShawnFumo Год назад
Well, there is a lot of videos on RU-vid and various discords to watch and ask questions. And Leonardo and Mage (and MidJourney, though not Stable Diffusion based) let you browse what other people have made and see the prompts and image masks and stuff. It sounds like the sponsor of this video even made a course. So there's a lot of resources out there if you look around.
@nikki-zeero3122
@nikki-zeero3122 Год назад
Lmao
@buffed103
@buffed103 Год назад
Music in bg?
@AbeOfArabia
@AbeOfArabia Год назад
incredible!
@parsleyisthicc
@parsleyisthicc Год назад
Hooray for laziness
@IPete2
@IPete2 Год назад
It is astonishing where the AI journey is taking us!
@max-or4lk
@max-or4lk Год назад
a is for the algorithm
@marverickmercer1968
@marverickmercer1968 Год назад
With everything available at a press of a button come the death of all meaning and value. The market for art, game, movie and all form of media and entertainment will become 0. If everyone can generate decades of entertainment for themselves with a button press, why buy anything from any one any more? Eventually everyone will just flood the world with their AI creation all day every day until it render human copyright meaningless due to the sheer volume of stuff generated, as anything new will be instantly copied and iterate on at speed beyond human comprehension. You will lose the right to your own creative work before it is even published. At that point why even bother ?
@dibbidydoo4318
@dibbidydoo4318 Год назад
if the meaning comes from scarcity, did it really have a meaning in the first place?
@marverickmercer1968
@marverickmercer1968 Год назад
@@dibbidydoo4318 Meaning and value come from quality. Imagine there is a game, a book or a movie you like. There maybe many thing you like or dislike, many thing that could have been done better, but overall it is a good and original work. Now do you like watching or playing blatant copycats of said work ? Do you play one game and then endlessly consume clones of it? Do you recommend your friend the original work or the copycat ? Now imagine an AI endless create derivatives work, each one changing a little but overall it just putting on a different coat of paints. The name may change, the graphic may look HD, but something still turn you off from them. And it took no effort whatsoever to create those derivative. Now, do the copycat still have the same value as the original ? Nowadays there is no shortage of anything on the Internet. Even before AI, people create OCs and write fanfiction and derivative works. Still we hold some work to be of higher quality and value than other. And we especially look down on plagiarism and copycat. Why ? I think the question answer itself.
@dibbidydoo4318
@dibbidydoo4318 Год назад
@@marverickmercer1968 if it's all a derivative, how will art lose meaning then? It will just increase the value of unique works.
@botkarma5753
@botkarma5753 Год назад
@@dibbidydoo4318 I think the point is more towards analysis paralysis and how over saturating the creative market (More than it already is) will make it harder for quality works to actually come above the trillions of mediocre pieces.
@BrgArt
@BrgArt Год назад
@@dibbidydoo4318 except it's not "unique" work. it's derivative meaningless big tiddy goth girl nonsense.
@Harsh1650
@Harsh1650 Год назад
OK, but can it replicate artist like MikaEmori, Iwa_to_mushi etc, who does different kind of colouring
@boomleon7224
@boomleon7224 Год назад
From what I know, many AIs will use the style of a specific artist if you name it, but I didn't try with this one.
@Harsh1650
@Harsh1650 Год назад
@@boomleon7224 both of those artists and a lot of others don't allow their art to be used for AI learning, I don't think it will be ethically possible to generate from those artist
@boomleon7224
@boomleon7224 Год назад
@@Harsh1650 idk, probably since it's posted on internet, they give the authorization of anyone to see it and take inspiration from it, AI included. That's my very best guess.
@taihuynhuc3135
@taihuynhuc3135 Год назад
@@Harsh1650 The problem is that it is super easy to download that artist's images from the Internet then train on them
@TavishVT
@TavishVT Год назад
​@@boomleon7224 if the artist does not want it trained, it is impossible to ethically sidestep it by blaming the AI for your own actions.
@Marune
@Marune Год назад
>ethical concerns lmao
@dee6561
@dee6561 Год назад
I really don't want to fall behind in the long run, but I also really dislike the ethical nature of this... What should I do...?
@Krishna-yi1dm
@Krishna-yi1dm Год назад
Bro iam using this in mobile
@tonynugget7223
@tonynugget7223 Год назад
Imagine all manga coloured at the speed of AI. But the art Luddites keep crying that work gets done too fast.
@cesar4729
@cesar4729 Год назад
It doesn't matter how many millions it benefits as long as it's not good for them individually. They like to call themselves artists, but "average human" suits them better.
@elfaranelasul6886
@elfaranelasul6886 Год назад
then make ur own manga.....
@tonynugget7223
@tonynugget7223 Год назад
Just imagine a handful of people competing with big name studios, pushing out content with a 1000 times more creativity like what we see with indie games. AI is a force multiplier and I'm excited to see the kinds of visions people have when grunt work and crazy costs are removed. An explosion in creativity is comming.
@nikki-zeero3122
@nikki-zeero3122 Год назад
@Arno Wisp yall dont even KNOW what the term luddite means and its funny really . In history , the "luddite" people werent mad because of "oh no technological advances!!!'' They were mad because of what people and their bosses did with those advances : fired them without a second thought, saw them as objects to be replaced , and underpayed+mistreated the still employed ones horribly . Much like what you guys are doing really - half of the "ai bros" advocate : You "artists" are done for! You have no power and you are nothing agaisnt this new fart generic thing we have now ! Your skills dont matter and every single one of you will die unemployed! ( and some going as far as to tell artists to k1ll themselves for it, haha .) Please do your research before deciding on a view ,as you tell artists' to do when they spill some logic to your ears about this ethical issue. Because it seems like you guys have already hit the "the ai will make us forget how to learn and think" step . Its embarrasing for everyone.
@nikki-zeero3122
@nikki-zeero3122 Год назад
@@tonynugget7223 and that all that ties to your unrealistically optimistic view . What do you think will happen when greedy corporations and bosses get their hands on this tool ? The only thing that will be exploding is the unemployment of their workers , after that moment. Just sacrificing the poor workers , to get a little bit richer , a little bit faster . This isnt a tool for creativity just because it makes things zero effort for non-artists . An artist paints and draws because they want to , the process of art isnt a hustle or a " meanningless waste of time wish someone else could do for them" . It IS for someone else though . Someone that isnt really an artist- just needs the results. Not the proccess , not the experience , not the learning , not the creativty. And that "someone" is capitalism addict , selfish bosses that try to blur the line of ethical work rules and pay , all the time . This is automation , and we all know, artist or not , that when companies get a hold of automation, things get ugly and not so creative nor passionate , very *VERY* fast . It wont be ending good for ANYONE but the rich corporate bosses .
@DSJOfficial94
@DSJOfficial94 Год назад
cool very cool
@g-ray4088
@g-ray4088 Год назад
screw it, make an ai that makes a text description of a image so we can have ai gartic phone
@The-Faraon
@The-Faraon Год назад
Could you use an AI like Whisper to place English subtitles, so the automatic translation is better? Thank you. Traslated by google. xD
@bycloudAI
@bycloudAI Год назад
ya was thinking of doing that, will add when i have time, thanks for the suggestion
@KurtStaInes
@KurtStaInes Год назад
Your building hype already this has not been released the v5 is just a preview
@samizo5842
@samizo5842 Год назад
Finally
@NiaArifah-br6cr
@NiaArifah-br6cr Год назад
AI prove that innovation always driven by coomer
@timaka46
@timaka46 Год назад
Common SD W
@yadav-r
@yadav-r Год назад
Whoa......
Далее
This Free Image AI Is Gonna Break the Internet
10:52
Просмотров 33 тыс.
Why Are Open Source Alternatives So Bad?
13:06
Просмотров 609 тыс.
Why "Grokking" AI Would Be A Key To AGI
13:36
Просмотров 60 тыс.
I THINK SHINY HUNTING SUCKS
11:23
Просмотров 1,5 тыс.
AI Generated Videos Are Getting Out of Hand
20:32
Просмотров 109 тыс.
The search for the biggest shape in the universe.
18:49
Next level AI art Control | My workflow
23:02
Просмотров 97 тыс.
I Spent 150h+ Making My Own Anime With AI
13:50
Просмотров 113 тыс.
AI Generated Art Is Getting Out of Hand
10:02
Просмотров 1 млн
How Stable Diffusion Works (AI Image Generation)
30:21
Просмотров 144 тыс.