These TikTokers Are AI Generated [SD Multi-Frame Rendering]
ฝัง
- เผยแพร่เมื่อ 20 มิ.ย. 2024
- 🕹 Get a browser that’s literally better at everything: operagx.gg/bycloud2 Sponsored by Opera GX!
When AI Influencers & Img2Img Video Generation meets each other... Literally everyone can run an model influencer account or even business in the near future.
National Geographic: How can you tell if a photo is AI generated?
[Article] www.nationalgeographic.com/sc...
TemporalKit
[GitHub] github.com/CiaraStrawberry/Te...
Hatsune Miku Workflow
[Reddit] www.reddit.com/r/StableDiffus...
Multi-frame rendering
[GitHub] github.com/OedoSoldier/sd-web...
[A random tutorial] • Video To Anime - Gener...
Mov2mov
[GitHub] github.com/Scholar01/sd-webui...
This video is supported by the kind Patrons & TH-cam Members:
🙏Andrew Lescelius, Chris LeDoux, Alex Maurice, Tony Jimenez, Panther Modern, Jake Disco, Demilson Quintao, 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, Brigham Hall, Kyle Hudson, Kalila, Jef Come, Jvari Williams
[Discord] / discord
[Twitter] / bycloudai
[Patreon] / bycloud
[Music] SlipStream
[Profile & Banner Art] / pygm7
0:00 Intro
0:49 Ad
2:15 img2img t2v - วิทยาศาสตร์และเทคโนโลยี
![CAMPปลิ้น | EP.76[2/2] เดินมาสิบกว่าป่า วันนี้รู้แล้วว่าโจโฉกลัวอะไร!](http://i.ytimg.com/vi/R7Yor-t2Au0/mqdefault.jpg)
Get another browser that’s literally better at everything: operagx.gg/bycloud2 Sponsored by Opera GX!
P.S. Looking for a video editor! DM me on Discord if you're interested
gi ve it a year and you gonna have custom ai anime movies
bycloud and the art of thumbnails
Opera is bloated asf
ai generated videos also mean ai generation gore and porn
firefox better opera gx is chinese spyware
whatever the hell im watching in my 40's gonna be something else
How old are you right now, comrade?
@@jankfranko488 39
‘the normies’ okay… you just lost me.
in my 40s I'm gonna feel younger than I am currently
@@DavidsuperZhong maybe you will be younger biologically if the current trajectory is maintained, DNA slash protein generation is going strong
Remember that this is the worst it will ever look going forward. Each iteration that noticeably improves quality is that much more likely to convince investors to jump on the bandwagon, so we are in an exponential growth phase.
What a time to be alive!
5-10 years from now I won't know whos real and who isn't by just watching a video.
To be fair, girls dancing are an excellent way to benchmark a video filtering algorythm:
- you have plenty of material to work with without much of a risk of problems with copyrights;
- they move in very unpredictable ways that puts the AI hard at work;
- because you're dealing with a subject everyone is familiar with, errors are much easier to notice and then worked on;
and slight imperfections, especially with the face and hands are VERY easily noticeable by other people.
you make some interesting points about using girls dancing as a benchmark for video filtering algorithms. It's true that this kind of content provides a substantial amount of freely available material for testing and refining AI algorithms. Their unpredictable movements do indeed challenge the AI's capabilities, making it a valuable testbed for improving accuracy.
Furthermore, working with a subject as widely recognized as dancing helps highlight errors more easily, fostering continuous improvement. Benchmarking against familiar content like this can ultimately lead to better video filtering and content recognition algorithms, which is crucial in the context of platforms like TH-cam.
Your observation sheds light on the practical considerations in AI development, especially in the ever-evolving landscape of online content. Thanks for sharing your perspective! 👏 #AI #VideoFiltering #AlgorithmBenchmarking
= This is not chatgpt.
Imagine if all this is applied to wonder dynamics. Heck if that happens then it's going to be completely mindblowing. Perfect VFX animations for everyone.
naw, we are still too far from perfect vfx animations.
gonna try.
@@arakemi1080 i think we'll see many mind blowing stuff this decade, made in someone's bedroom using just a laptop lol
@@SeaSerpentLevi yeah, this decade, but we are still far from perfect
Time flies, a decade can look like much, but really it’s nothing, it will go and you won’t even notice
"this is fine" meme is a perfect fit for our future. This legit terrifies and excites me at the same time.
I can easily imagine this being able to run realtime in a few years and you being able to video chat with fictional characters. Pygmalion is already a trip, being able to talk to them semi-convincingly.
I want to be realistic furry
can't wait to date anime girls that won't be real yet
@@agentbanana0766 The future is now haha
@@agentbanana0766 they will also get disgusted by us 😔
There's already a guy running his Pygmalion AI bot on twitch as a streamer, and yes it can respond in real time (dev name is Pladis).
This takes deepfakes to a whole new level. Literally rebuilding the entire person.
Her. We can remake Her.
@@arnowisp6244 We have the technology.
Yes, thats what AI does. We all knew this long ago. why the f are people so shocked now?
this tech was around before it was consumer available.
now think of all the deep fakes you've watched in your lifetimes, and never even knew it.
This was a movie once.
Turnes an ai program into a video chat site. Was so realistic the ai turned sentient and had genuine feelings and did art and poetry themself.
Wow, I can see videogame studios upscaling their old franchises this way.
The techniques shown here only work for videos. If you wanted to use them while moving around in a game yourself, you'd have to generate those images in real time, which isn't feasible with current hardware. I bet we'll see more promising approaches where AI modifies polygon models, textures and perhaps animations soon though.
@@martinm6368 I mean, they can upscale their characters and textures by keeping the same architecture. Certainly, your idea of real-time upscaling will be more challenging to make.
@@DJVARAO I expect projects which process old games like that soon. However in the clips shown here they've processed a recording of the game. So it's just another video.
@@martinm6368 it's not feasible with current algorithms and hardware combination. but i've seen some image generation algos doing 10 fps with current hardware. that's inching on real time, and it's just one paper down the line. who knows what we'll have in two years.
As an illustrator and animator, I for look foraward to what Ai is going to be able to do. Imagine what the implications will be for video games and anime. If someone like me teams up with a tech nerd who knows how to use Ai, i think I can achieve my creative dreams even if I don't have a massive budget.
@Rivershield agreed, I see AI lately to enable smaller grousp to make images rather than a whole FX Team to build and render the images. I am not sure the impact yet for games as there is still a need to build levels and interactivity but I am sure AI will be able to generate gameplay and robolox / Minecraft like stuff soon with little programming....
Problem is, soon a tech nerd is not gonna need your skills.
@@morosov4595 That Depends if you can show why the AI can't Design as well as you.
@@CGBOSSS Also, storyboarding would still remain an issue. There's only so much an AI can do for now, so it might be a long while before we see a "good" anime only done by tech nerds.
He is an illustrator and animator, not designer or writer guys.
We'll be getting to a point where creative work has been influenced or taken over by AI so much it'll become derivative as a whole, and then we will have to invent something completely new to go beyond. Either that or we'll be stuck with seeing every day infinite reiterations of the same faces, animes, artworks, games over and over.
The same thing over and over? Just like a Disney move, honestly.
@@ClockworkGearhead ye we've already been stuck with the same AAA games being re-released every year as well.
@@chickenmadness1732 YA'LL LIKE ANOTHER CALL OF DUTY!
THIS TIME WITH AI GENERATED ANIME DANCING GIRLS!!!
the next art form will be one where instead of physically creating music or painting/drawing themselves, the next generation of artists will instead program ai models that produce unique artwork
We are already on that, but it's just that it is not done by AI, AI will just make it easier in the long term.
If you're scared of AI, always have a mindset like this.
"Treat them as your tool, not as enemy"
I also use this mindset if I have useless member in my group.
Yeah. An insanely powerful tool that will be weaponized.
But I'm all for it. Rushing headlong may be our only chance against climate change anyway.
And if we are doomed regardless, at least with AI we may die alongside our personal waifu. Worth it.
I mean, if you're in the entertainment industry, making tiktoks or even creating anime and stuff, this ai literally is your enemy. You're being out-performed and competed out of your market by these bots
FINJALLY A SENSIBLE OPINION ABOUT AI
well that's an extremely retarded take...
AIs feel the same about you :o)
"Days before them" You real hit the nail on the head with this one. This Open Source approach makes things develop at near singularity pace.
AI art is more real than most girls Tinder profiles!
gotta agree with you on that, tinder girls are more fake than AI generated images
more real than the height men list themselves as on tinder
@@nicholash1278 I’m actually a 6’3” Asian man, trust me.
Not really, the AI girl still looks like unrealistic anime. Maybe in the future people will value flaws again to differentiate between real human photos and AI generated portrait.
@@nicholash1278 more real than you
Text to Video AI is really crazy, and I think it would just be really fun to make random AI generated videos once the technology is mainstream imo.
At some point in the future assuming some technical limitation not yet understood doesn't stop the progress of this tech.. its going to annihilate the film and TV industry as we know it..
@@johntowers1213 Yeah, it's going to be really interesting to see how this plays out in the future.
once? you can literally make them today. What is holding you back.
@@nicholash1278 it looks like dog water
Once text to video becomes better we all know what key words people will use to make them.
"Busty asian woman, cat ears, sitting on the bed, spread legs, touching her lewd parts of the body" etc...
The most popular searches regarding AI are AI generated girls, how to ERP with ChatGPT and AI generated porn.
Holy cow, the AI-Generated videos are getting good, specially the lighting and shadows in cel shading.
Everyday I feel like I should yell “don’t trust anything you see on a screen” louder and louder
THE THING IS:
We just started with AI, it's scary how good this is.
There will be a time that you won't be able to know if it's a full computer-generated video or not.
We're shockingly close to that already. It makes me fearful that we won't know what is real and what's not. How can we choose a good president/prime minister if we don't know if the video evidence of their crimes are real?
we haven't "just started" AI has been around for 25 years, its just not accessable to the public.
@@mis8866 25 years is like nothing.
@@gabrielnilo6101 Nothing compared to what? 25 years is a long time. AI isn't "just started" it's only just been readily released to the public
Too much of a good thing can be bad. Eventually technology will come to a certain point in history where its no longer good for the mankind. I think thats not too far away now.
Just generate a normal image from the first image and use a style transfer, they work pretty well
As an artist this is so damn depressing, but I didn’t realize just how much I wanted to see old anime redone with AI to look like live action… will someone do all of Akita for me?
why is it depressing? there was already an over saturation of art and artists that had almost zero demand
Physical art will always be valued, but digital art is going to be killed by AI. You can still be a digital artist, but your skills in digital painting will be worthless, and instead your value will be in how you can compose an image with AI.
@@truthdoesnotexist that's really just saying that there was a lot of water in the lake and only a few fish. (water=artist/art, fish the demand for artist)
while this A.I stuff is an ocean of possible content oversaturation becomes an understatement when the skilless are granted the ability of the skilled.
I'd be depressed too if I was a measley bottle of water going to join the pond but then the fish start making their own water. thus making my already unnoticeable existence abysmal in The grand scheme of things. I'd rather be a cup of water in a pond than an ocean.
@@__aceofspades Yeah I agree, But good art fundamentals transfer to any medium, especially your visual library. An experienced artist who knows how to use AI will almost always do better than someone who only knows how to use AI. It's like how photographers take better pictures than people who just have a good camera.
I agree that technical skills for painting software are not as valuable right now, but I don't think we should completely write them off. With new technology like augmentation, those skills will be useful again. (Not like those crappy scribble-to-art things)
Anyway, I don't think traditional digital artists will disappear completely, at least not in big studios. AI-generated art can't have copyright protection, so studios won't want to risk that. But in the freelance/indie world, yeah, traditional digital artists might fade out a bit. as a digital artist for almost 3 years it Fking sucks having my skills degrade in value, but im also happy that we're advancing art in general.
@@YeviCoulson well their not really skilless, they are more technical minded people. its not as simple as entering prompts, you have to learn how to communicate with the AI which is not as easy as it seems and takes a technically skilled person. Someones grandmother is not going to be making AI art just yet. And the art world was already completely over saturated already by basically unwanted art, it was mostly trash anyway. Now artists will finally have to start getting creative and make things that are unique. And AI is just a tool for real artists to use. My dad is an artist and he uses AI generated images for textures in digital art and in collages. If your an artist and scared by AI then you probably had no real talent in the first place. hyper realistic images are somewhat impressive and take skill, but have been obsolete since the invention of the camera, 3D animation and digital tools like Photoshop. Not to mention hyper realistic portraits aren't unique.
Honestly not surprised by txt2video for now. And I don't mean that it sucks, I mean that AI tech has become so good that I have raised my expectations even higher than that! I don't think people will approve just taking a reference and sticking a filter on it, no matter how good the result. AI needs to be able to generate videos by itself. But it's a great start, as we now have tools to keep consistency between pics (which must be great for image generation too!)
The good part about references is that it is one way to tell the computer what we want, that we can control pretty well. Prompt generation can be a bit random if you are not incredibly detailed, but I agree, it will be very interesting to see how prompt-to-video generation develops.
Nvidia's paper tamking about temporal denoising can be applied in a very good way for this ! I'll try to code it into automatic1111 but I'm bad at coding, and I would have to finetune the model with videos after adding the time dimension, and train the temporal alignment model... They should just release everything as open source already
you got a link to that paper?
@@dobaovuongtamdamkt4197 there are several ones but the latest that really interested me was (I can't send links so I'll send the name) "Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models"
@@Kynatosh omg so new, Karol released a video about it only 5 days ago
I tried doing variations of these, multi-frame, tempkit, ebsynth. So far, it's easy to only change face, or slightly different clothings. However, it's just very difficult for me to make, for example, dancing video based on some other dance video, with my own lora trained character with different hairstyles, clothes, and accessories.
The Cowboy Beboy anime-to-RL part looks awesome! Something something funny quip about Netflix Cowboy Bebop...
Holy crap dude, I never even realized that AI Image generation could, in theory, be applied in real time to game software. In fact, with my experience with shaders, you might be able to make any game hyper-realistic if we could train an AI model on transferring a game's graphics into real-life graphics and then overlaying it on top of the preexisting game. This could actually revolutionize game graphics and I realize that I might be giving away a billion-dollar idea here.
The main problems with this I guess are the still lacking object permanence that comes with ai image generation models, as well as the render time being nowhere near real-time, especially if we want to get "realistic" results. There's also the fact that you've got to somehow find a way to train the model you're using to not render the GUI as something that isn't the GUI, as playing an FPS without a reticle, for example, would hamper the game experience past what the graphics could provide lmao.
That's why it would have to be built into the render pipeline so it's done as a post process after the 3d rasterization, similar to anti aliasing. Reshade can do this, so there's no reason an ai filter pass couldn't make its way into Reshade or something similar.
Intel developed something similar to this for gta 5 with a shader trained on German dash cam footage. This was a really long time ago, so i wonder how good it would be now 🤔
Nvidia already has DLSS. It's only a matter of time for them to make a more complex version of that model that actually transforms the graphics instead of only upscaling them
Chasing realism is cancer.
@@mauree1618 I’m not suggesting this be used as the “last frontier towards achieved realism” or something. Just as this can be used to create ultra realistic environments and people, it can also be used in a near infinite amount of ways in order to perceive whatever graphic style you could ever want.
Imagine being able to write a prompt, just as you are able to now with chatGPT and Midjourney, and be able to experience any game of your choosing through an entirely new, personally curated lense. The implications are as awesome as they are frightening.
Who knew that even when it comes to showing off new tech, thirst traps always work. No wonder why these videos are so popular on Platforms such as TikTok.
Fantastic video. Gonna try diving into this today to see if we can demo this next week on our show.
Might need a new GFX card... AGAIN
I happily accepted the new waifu overlords!
Embrace the chaos. Accept it and enjoy it.
"enjoy it" sus choice of words.
@@bazookaman1353 don't kink shame
@@lpnp9477 I have loli content on my phone.
@@bazookaman1353 god bless
One thing I want to try (but don't know how to code cause I'm a dumbass) is 2img2img.
Similar to img2img, it takes in one image, does whatever the hell to it that img2img does, however, it takes in a second image as a sort of "control"
It'll use the control to guide the img2img to retain features between frames.
For example, the spinning house, it just generates whatever so doors appear and dissappear whenever they feel like it.
But using 2img2img, the control will give it the information of where the door should be, and how the door should look, while also knowing where the door should be from the input.
I'm probably describing something that already exists but idk. Maybe that's what ControlNet is. I don't fully understand controlnet
The moment you will be able to have a filter like this in a game @_@ and literally make any 'old' game look new and refreshing, by pre-rendering future FPS. Either way, it seems that technology will go in a cool direction in the next decade, wish I was born later to get this technology already + even way more insane technologies that there will be.
The irony I think is if you were born later you would appreciate it less. People take for granted almost everything familiar.
Bruh imagine intentionally developing a game with shit models so you can get it to market faster and then just giving it an AI makeover as a final touchup before launch 🤣
@@theholyjosh5384 not a bad idea, especially if you don't like making models
The leather jacket in that Cowboy Bebop fight scene, it looks pretty good in some of the shots!
Used to be "internet is for porn". Now it's "AI is for porn", and I'm all for it.
me clicking a new bycloud video with my meat in my hand: Sensational...
wow bro, what's your method to get zll these videos to use for your edited video?
Them tiktok videos that is AI generated video looks almost real and just take it a lot of months or even a year for this to look more real and amazing video bycloud :]
It didn't take them "an year", the whole thing started around September when Stable Diffusion arrived. Also these videos are relatively recent and they all seem to use Controlnet, which means they are pretty new.
3:55 "even oversaturate" that is putting it veeery mildly 😅
there is absolutely no reason anymore to pay a real model for a photoshoot. like it's completely pointless now.
There are pros and cons to this.
People don't need much makeup if at all to generate the videos.
But the watchers may grow even more insecure about their beauty. Hard to compare to AI.
IT KEEPS HAPPENING
thanks as always for the super interesting breakdown
That cowboy beebop clip was nutso.
I'm just confused why is always an Asian woman reference video turned into a slightly prettier Asian woman with AI. Like if you're trying to show the power of your kit, turn an Asian woman into a black man or something. This just makes me think the tools are extremely limited.
Because most people working on tech are Asian men, and what they do with it reflect their tastes and preferences
I prefer my women without a 🍆, but that’s just me.
Be the change you want to see in the world. Should generate a lot of upvotes.
I fully embrace AI tech going into Anime. 95 % of all studios haven't progressed in their animation or even dumbed it down further visually. The only studios pulling their weight are Ufotable or Mappa it seems.
Not really, there are other good studios too, A1 for example
That's very true. One of the reasons One Punch Man was so good was the badass animations early on, but they got spotty at other times. I guess it's still a hard process so hopefully AI can help them boost quality in future.
not really, not everyone is a fan of western artstyles, I personally have PURE HATRED towards shounen art styles like one punch man, mob psycho, my hero etc
@@ThePr0Br0"pure hatred" lmao, ok, what? Did it give you cancer?
This video is such an asset for noob me, I'll be keeping the link in my notes, :)
Also I am an old subscriber, have been subscribed for 6 months or so
7:33 LOL DID NOT EXPECT TO SEE THE KAMIKATSU SCENE HERE
Damn that's impressive. Got a few v100s gathering dust if anyone wants to remote in and tune out some stuff. 😂
Haha dude I still love you :D
Hey sex sells...it's just a reminder how basic we are as human beings no matter what cool stuff we're doing 😂 Accept it, embrace it ❤ love the content!
And when people make deepfakes nudes of in real life people without their consent?
@@zkme2734 sorry I was talking about the promoting AI research by using dancing anime girls. You're talking about something else I think. Did I miss this in the video?
@@madsmadsoleh8642 I think his mindset is more like, 'I can't believe this technology exists,' so he takes the worst thing that exists with it and applies it to every video he watches, even if it doesn't make sense in the context.
@@mizuki4155 funny how the worst thing is always humans being basic 🤭 if you need deepfakes of anyone real, you are seriously limited by your imagination 😅
@@mizuki4155 I mean, he said "Sex Sells" they're not wrong to assume that.
4:26 Holy s**t now this is creatively more intriguing and unexpected!
That dance at 5:53 is so amazing.... love it so much
where can find the dance source video
Let me just say something about AI generated art and animation. To those who sit around doing nothing will still generate NOTHING. In reality, with or without AI, the process of generating a COMPLETE piece of creative content is still really hard. Creative and problem solvers are still gonna be the go getters while others are still you know sitting around
My man, you speak too much sense. and for that no one on the internet will take your advice seriously lol
@garen591 Yes, this is (partially) true. It's similar to how there have always been attractive women and women willing to get undressed yet only a tiny minority of them are profitable on OF. Additionally prn is technically free yet many people still buy stuff. I like your take although I do think there will be a considerable impact on novice and midtier artists.
well there is gpt 4 and gpt 5 coming which are insane at creativity. and the methods to harvest that creativity are slowly coming
@@throwacnt7603 I agree its only partially true. I mean way less girls were selling nudes before the days of OnlyFans thats for sure. Also look at music production so many artists just come out of their bedrooms. That just wasn't a thing 30 years ago way more people just didn't create because the tools were harder to use and access
@@dawre3124 gpt 4 is trash at creativity. it's still at party trick level at best. for it to generate anything worth using, it has to be babysat pretty hard, to the point you might as well do it yourself.
i'm a 3d artist i'm not a good illustrator so the thing is if i can get hand drawn result by my 3d model without spending hours to generate shaders and doing postwork i'd be really happy.. thing is they have some of those limitations in control
7:35 I think the AI may have been a meta joke considering the caption “Way faster than doing it by hand!” Lol
your explaining things very well, thanks bro, and I like your accent
LOL if people can't see these faces are fake, ppl need to check in with the doctor.
This is the worst it will ever be.
OF girls are complaining of decreasing revenue as text to image models improve and can be run on more commonplace hardware.
It is unfortunate when jobs are replaced by technology, but it is a symptom of progress. I still don't know what I'm going to do exactly when programmers are replaced but there's nothing that can be done to stop it so maybe I can start an OF idk
@@lpnp9477 OF is already on its way out. There will be new jobs related to AI but a lot fewer that those that will disappear. At some point, we will need a new social model or there will be more chaos caused by idle hungry people.
@@Kevin-jc1fx are you telling me no one will pay for nudes of a male ex programmer desperate for food and shelter? Oh gee
i am so excited to see what AI has to offer in the future as I also learning AI programming.
It is all going back to a cyberdeck and scripts. Can you imagine DVD's in 20 years with Scripts?
Its painful to read reddit where some people bash AI videos as being terrible and hopeless, yes they arent good TODAY, but like images and text, we will see huge improvements in the upcoming months and years. This is still the infancy of AI, and especially video AI.
it's just people with zero foresight. they'll be eating their words soon enough.
Can't wait when i finally will be able to print my personal catgirl
Holy cow! I'd rewatch watch Cowboy Bebop remade like in your example!!!
Imagine when AAA game companies can just get an AI to shit out a derivative game annually just by updating a few name tables.
Bruh people saying this is cringe are so single minded... This will revolutionize all the entertainment industry holly sht!
woman are in trouble very bad news
dw we like it too
The real trouble is when they create r18 deepfakes of them
Wait until the male ones come out
At least we won't see as many people simping online.
Women will be fine, simps will lose it
Been having a lot of fun with this tech, its only gonna get better and better!
is there any tutorial to produce such AI video, IM totally new to this, do you have any relevant past video that I should start off with. Thanks for teh vid
im excited by the proposition of more high quality anime coming out by having it do all the filler frames
I had to subscribe because that guy in the back ground told me to during the better version of the fight from Cowboy Bebop.
The anime one made me laugh out loud last week. It just came out of nowhere.
This has to be the most persistent recommendation of the year! Fine, you win! I'll watch it😂
The point about games is interesting, what I am really waiting for is live display editing/censoring vidangel style, where you can describe what you want to be different about say a character in a game or show, and whenever they appear on screen the GPU will make those changes before rendering.
So if you wanted to deepfake Rock the Dwayne Johnson onto Jar Jar Binks, or vice versa, you could watch those unedited movies on a streaming service, but on your screen the changes would be applied.
haha the ending is really funny 😂😂
Is there an audio version? I was thinking of doing this to self dub anime XD
The prevalence of anime-related/styled content online has a lot to do with engagement-based metrics and an audience that feels like it has to be there for everything that happens, even if it’s in a country 12+ hours ahead.
You're right about that, I never do dancing girls or hot girls and no one sees my animations lol
the live action beebop looks hilarious 😆
sick! The Live action cowboy bebop! Can't wait for this technology to start pumping out remastered/livefake oldies. Old Video games, older movies, remastered!
I found another way to reach consistency and avoid flickering
Could it actually be possible to give a rigged model to an a.i and have them use that as a reference when generating frames for characters?. And, for added consistency for rendering those animations, the A.I should be able to first, generate that character in full using the character model as a reference, store the data and then use it to output the animation. Imagine a model sketch from every direction and angle.
Like, would it be possible for the a.i to first render every element in full before hand and you can either delete it or store it indefinitely after use?
That would be really cool.
imagine the future of cinematography...you put some seed story paragraph into a text generation AI linked with some text-to-video AI, and boom you get an anime episode!
The only issue now to have a way to let AI innovate new art/story styles. The newly created art style can then be put into a generative agent society that models the current human society of the geo-region of intended release to test out its popularity and rate of organic spread. This cycle is performed automatically until it finds out the most generally liked anime episode.
cinematography refers to how film is framed, like camera angles and movement. That is not what you are describing. And 'the most generally liked anime episode' sounds dreadful because by definition it would be the most middlebrow end product possible, MCU-tier. A lot of the best anime (anything) is divisive to at least portions of potential viewers.
Are you telling me that the POPE don't have drip?!
What is making these ai picture dance called? Cause I wanna make it
Just when I was starting to think technology was stagnating in recent years, we get a whole new branch off that will change so many things! Most excited I've been about tech news this last decade easily.
to me this is the most exciting era since the industrial revolution. people have no concept of how big AI is yet. everything is going to see exponential growth in the next years. maybe i'll finally get my anti-aging drug that i was promised back in the 90's, that my generation would be able to live young for their entire lives.
Yeah I'mma need that VTM mod asap.
There's no stopping this rate of improvement..just imagine how different the world will be in 2030
that person did wispered in my mind to sub. U got me there
The future is Netflix type of service where you write a prompt of what kind of movie you want: Horror pink alien movie 30% comedy in a ciberpunk planet 10% musical with brad Pitt style main character…. Enter.
Movie a la carte. No director, no actors, no script writers…. All AI
Im still waiting for the technology from that Black Mirror episode "Be Right Back", literally making an AI generated person "real". We are getting very close to that point, at least in the generating face and body part of this whole thing
Lmao remember how much money they spent making those anime eyes for Alita Battle Angel? That can be reproduced by one guy with a 4090 now
2:44 coherent? Not the word. What you're looking for is "temporaly stable"
Damn, AI growth has been quite insane these couple year
*months - literally in a few monts the landscape has changed drastically, stable diffusion is less than 1 year our
Basically actual terminators of skynet would be dancing anime girls
what subreddits are good for viewing more of these videos?
The AI jank kinda works for Masquerade
Can link to the dancing videos?
good video, would be best if you linked the videos used in it.
Lol that's why that anime scene looked so weird 🤣
Rock Paper Scissors Anime was pretty good no lie. Great content here :)
They will combine this tech with what they used on Mandalorian.
this tech was around before it was consumer available.
now think of all the deep fakes you've watched in your lifetimes, and never even knew it.
things are moving so quickly. we will soon have flawless videos made with AI!
pepsi man !! now theres a blast from the past haha