When AI Generated Videos Meet The Uncanny Valley
Vložit
- čas přidán 7. 04. 2023
- Check out today's sponsor RunPod to get started on the best priced GPU cloud services using this link: l.linklyhq.com/l/1jHv6
Text to video generation is at the stage of dall-e mini for text to image generation 2 years ago. We are about to see some really interesting development in the coming year.
Text-to-video-synthesis Model in Open Domain (Shutterstock T2V)
[Model Scope] modelscope.cn/models/damo/tex...
[HuggingFace] huggingface.co/damo-vilab/mod...
[HuggingFace Demo] huggingface.co/spaces/damo-vi...
[Damo Lab Webiste] damo.alibaba.com/?lang=zh
Runway Gen-2 Demo
[Link] runwayml/status/1...
Will Smith Eating Spaghetti
[Link] altryne/status/16...
Elon Eating Spaghetti
[Link] abrahamstarosta/s...
Darth vader in walmart
[Link] / text_to_video_darth_va...
Text to Video Star Wars
[Link] victormustar/stat...
SD Grid Method
[Link] www.reddit.com/r/StableDiffus...
This video is supported by the kind Patrons & CZcams 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
[Discord] / discord
[Twitter] / bycloudai
[Patreon] / bycloud
[PayPal] paypal.me/bycloudai
[Profile & Banner Art] / pygm7
[Video Editor] @askejm - Věda a technologie
If you want to train, finetune or run models that needs high VRAM or computational costs,
check out today's sponsor RunPod to get started on the best priced GPU cloud services using this link: l.linklyhq.com/l/1jHv6
One day people will say things like "i remember when videos were real"
I mean... Films are CGI for quite a long time already
@@absolutehuman951 I think by "real", he meant that it was created by humans. In this case, videos were made by AI are bizzare.
I'll keep it in mind for when I become old and want to impress my kids. "You know, back in my days, people were real".
And no one will believe them
That’s now
All the AI generated videos have a weird 80s home video vibe.
The bright colors, the low resolution, the cheap-looking props.
....for now. Soon they’ll look 8k
Or like old 80s cartoons/animes with all their animation errors we forgave them but wouldn't forgive now.
Right
@@HypercatZ wtf anime back then looked good.
The early years of AI generated content needs to be recorded and preserved.
They will serve either as comedic value in a advanced future utopian society, or reflection on the harbinger of humanity's doom.
They will be like looking at the pixel art days of video games or the black and white days of movies.
the early years of text to image were very creepy -Human from 2040 that doesnt even relate to what i said..
@@perrolocoyt The most easiest and clearest answer why all religions are false is: ALL RELIGIOUS SOULS LIVE IN BABYLON.
CHRIST SAYS:
Revelation 18:4
And I heard another voice from heaven, saying, Come out of her, my people, that ye be not partakers of her sins, and that ye receive not of her plagues.
Muslims live in the babylon, being partakers of her sins.
Hindus live in the babylon, being partakers of her sins.
new agers live in the babylon, being partakers of her sins.
Atheists live in the babylon, being partakers of her sins.
Ignorats live in the babylon, being partakers of her sins.
Satanists live in the babylon, being partakers of her sins.
Catholics live in the babylon, being partakers of her sins.
Lukewarm christians live in the babylon, being partakers of her sins.
Christians = souls who are bible believers and obeyers, live being out from the Babylon and expose her wicked deeds.
Matthew 7:21
Not every one that saith unto me, Lord, Lord, shall enter into the kingdom of heaven; but he that doeth the will of my Father which is in heaven.
James 2:18 Yea, a man may say, Thou hast faith, and I have works: shew me thy faith without thy works, and I will shew thee my faith by my works.
James 2:14 What doth it profit, my brethren, though a man say he hath faith, and have not works? can faith save him?
2 Corinthians 13:5
Examine yourselves, whether ye be in the faith; prove your own selves. Know ye not your own selves, how that Jesus Christ is in you, except ye be reprobates?
Without the needed works, how can soul EXAMINE themselves and to see that they to be in Faith?
And if only FAITH is needed and not the doing of the WILL OF GOD (WORKS), then why there be only FEW who find the NARROW ROAD TO LIFE?
Why many perish, if many say they have faith?
Matthew 7:13
Enter ye in at the strait gate: for wide is the gate, and broad is the way, that leadeth to destruction, and many there be which go in thereat:
Matthew 7:14
Because strait is the gate, and narrow is the way, which leadeth unto life, and few there be that find it.
Only if the creationists dont nuke the planet
Ai is going to kill us by then
Things are moving so quickly it's staggering.
Yep
I am getting in love Ai now.. 😔
Fr.. this is scary
yeah in the past it tooks 100 years to invent new technology now it took only a year to invent a new technology
Too quickly
I want to see a VR horror game using these early "primitive" text to video style visuals. It would be absolutely nightmare inducing.
I remember pretty well the things I usually see when im on drugs, and holy shit, this looks exactly the same to the things I imagine during my trips. It feels so strange and uncorfortable to be able to see this videos without ingesting any kind of substance.
This speaks volumes.
Dude, for real. The fact that you can't make out most of the visual details reminds me of some of those overwhelmingly chaotic visuals you get on psychedelics. It's eerie and leaves me feeling perturbed.
@@Cedricson Well the way machine learning works is alternating linear and nonlinear transformations across `neurons` really nodes or dimensions of latent space. Basically though it is math.
a lot of people say that these images remind them of their dreams and trips. my analysis is that computers must really be thinking like human brains
ChatGPT and StableDiffusion are crazy, but I don’t really find them too scary. This however, really scares me
Especially when they eat spaghetti
awesome!
what is this program?
@@pigeonsil240sx I'd like to know as well. I know that Google has one but it's not been released to the public.
this technology will revolutionise filmmaking. it's great for storyboarding and draftung scripts. this is the 3d printer of the film industry
This AI stuff is getting too good so fast, I just watched the movie A.I. again the other day and I remembered the quote from gigalo Joe "They made us too smart, too quick and too many. We are suffering for the mistakes they made because when the end comes, all that will be left is us."
Hey I also watched it a few weeks ago. On tv. Don't like the end.
If you think about it, this is how our dreams look like. We are finally able to create videos of our dream descriptions. This is wild.
from what i know, EBsynth is not an ai tool - You paint the keyframes, but it generates frames inbetween, by breaking the painting into small shapes and then interoplating them. At least i heard something like that...
That is true. I don't know why everyone keeps thinking it's AI. It's just a very good optical flow algorithm.
@@drdesten It probably uses a neural network for the "breaking into small shapes" stage
Yes immense proportion! Makes you wonder how does they fit to pass the starship doors and load limits...
1:49
Fever dream Vader can't hurt you.
Fever-dream Vader: Y'all are fat.
:'(
They grow their warriors to immerse proportion 🤣🤣 I am dead
Surely they fuel their oversized defenders with these.
*immense 😅
Screw it, I'm gonna start eating spaghetti with my hands and eating it through my nose and eyes.
😂
1 year later this will improve greatly
It did
@@user-my3sp4oi4r yeah
Darth Vader fatshaming people of Wallmart made my day. Thank you so much!
Oh man, this one made me laugh with that Darth Vader video. Really impressive developments we are seeing. Thanks for sharing the news on this.
I like videos that mix humor with the serious to both entertain and teach people who don't understand these things well.
Oh my God the Darth Vader at Walmart was the funniest shit I’ve seen in a long time😂😂😂 2:38
that puffy jacket was fire, though. The Pope needs to request a real one.
"This primitive world, Earth. What do they have that could possibly stand against the Empire? This giant supply station must be an advanced military outpost. The Wall of Mart. Tah, I see. They grow their warriors to immense proportions. Most impressive. Perhaps we should recruit these giants for our own army... or not. As long as these battle a stockpiles of high-calorie sustenance. Surely they fuel their oversized defenders with these. I can't imagine the sugar rush they must experience on the battlefield. A whole section dedicated to miniature replicas of their heroes, how quaint. I wonder if they have a little _me._ Oh yes, they do, complete with a tiny choking action. These domestic contraptions appear to be more advanced than I expected."
I remember a comment on a different vid saying "we are in the silent film era of ai generated videos" and I couldn't agree more
So generated videos are so cool and nice video man :]
damn the future is going to be....Weird
Yep
I find many AI videos to be very surrealistic at times ... there are a few I actually like a lot thanks to their dreamlike environments. ""AI FILM -The Carnival of the Ages - Runway gen2"" is a good example.
"I wonder if they have a little Me? Ah yes. They do. Complete with a tiny choking action." 😂😂😂😅😅😢😢
Not sure if Corridor Crew used Ebsynth, They basically solved the flicker by adding images of the actors at various angles to the image dataset, it allowed the AI to keep the images of them consistent through all the frames.
Wow
Same time
DANGEROUS
The Darth Vader warrior bit was foul lmao
Ive been listening to the David Wong novels recently. Mixing cosmic horror and text to video AI seems pretty interesting.
I’m wondering what the creature descriptions would generate. I’m guessing it would be less terrifying that a person drinking beer or eating pizza nuggets.
i'll wait a hundred years for an ai that will make text to anime!!
Wait a year that's enough.
@@wolfganghorn7492 kinda of scary to know this is probably true
I will give it 20 years at most
I think u guys underestimate how hard it is to make a text-to-video AI. It will probably take 2-3 years minimum.
@@khytron06I'll wait even a hundred years if necessary.
This text video thing is very interesting, I might even be willing to pay a little for it.
Gracias por tanto trabajo de investigación. Que tengas buen fin de semana.
I realy like that , it's exaclty like a dream the objects change forms all the time, and the end of the dream have nothink to do with the begenning of the story lol
And this is just the beginning... Do those who already knew him remember Dall.E, the first one to appear? Yes, it wasn't bad, but it had a lot of glitches, problems, blurryness, etc... And that was just... 2 YEARS AGO! The same goes for Text2video... this is only the second generation and it already looks pretty good... I don't even want to imagine what this will be like in just a few months or 2/3 years in the future if it continues like this.
think about it though. being able to generate a Pixar film within a matter of seconds, or making the continuation of any movie you want, even making animes within the blink of an eye. the skies the limit.
I can't wait
Do you remember early text to image memes this feels exactly the same
Chinese bootleg dvds about to be lit!!!
Best channel on CZcams
Imagine if Google made a text to video AI that was trained on every CZcams video.
I want the option though to not have temporal coherence. Some of the images made are more cool than the first one made, and I would like to find the best one made then make that consistent.
My hunch is that text to video needs a kind of 3d latent space keeping track of the subjects.
WIth only 2d processing I can't imagine that there will be good coherence at all.
It doesn't even keep track of objects in 2D once they go behind other objects. It doesn't have layers.
2:20 the lower left of screen, clearly says "Star Wars" but as the camera moves, the words don't pan off the edge of the screen, so much as they get squashed into it and each letter becomes less legible.
These crazy videos look like dreams to me 😮
1:58 basically "I told AI to generate America" Yep..
This could easily be done if it includes agi,3d generetor. Agi will help to produce accurate things and 3d will help it to stay coherent
Ah yes. What a usesful information. Don't think researcher thought of that.
Just give the AI sentience and It'll generate video well.
(I'm kidding btw. 😅 )
@@poutineausyropderable7108 I mean we gonna go there one way or another. Gpt 5 may be an agi and gpt 6 will definitely become an agi. And we have plugins coming soon too,you may not even need agi for this just a 3d plugin plus video generetor plugin could be enough. Although our 3d generetors are still in its infant stage so...
Wow i think i'm liking the Ai now...
It's not just about 3D. Like how do real people move? They're in one pose, then another, they just flow. They don't just move each part at continuous speed between positions.
Where can we go to play with this?
This is going to shake up Hollywood. You’ll be able to prompt your own movie staring you and your date. Hell, you could even just add yourself to a movie at this point, with the right skill set, prompt, program, etc. 👍 The future is going to be wild
That’s what I always thought . Why would you watch the next marvel film, when you friend makes a Marvel vs DC crossover in his bedroom
@@miamitten1123 Exactly. I think the future is going to be something like this 👍 there still will be the need for good direction, acting, dialogue, maybe some editing. The AI will probably not be as creative as a person (at first) But those will be easy to solve eventually.
I want to make my own. Looks like fun tbh
They grow their warriors to immense proportions fucking lol
can anyone tell me what program this is or what ai will generate weird creepy things like this?
2:12 did darth vader just spawn out of nowhere :)
Hey I have an idea. Can you make a short analogue horror video with these A.I things?
too cursed? no such thing, I love the weird stuff.
When all civilization has gone this will get dug up from a time capsule by new in habitants saying "these humans were stupid"
1:56 the ai hit a liminal space
Yes the generations look like something from the movie "Waking Life."
that Darth Vader sounded just as much like Darth Vader as it did Dagoth Ur
I almost wish people would stop working on these for fear of what they will become
Thats INSANE!
Aesome!
What's the AI used to generate the images at 3:51 ?
midjourney
midjorney v5
Thanks!
Me : "Back in my days, digital media like films, digital arts, songs, pictures were all still handmade by people.. Back when we still need put our sweat and tears to mak-"
my grandkids : "You've said this multiple time already.. Please let us make video in peace.. I need to find ideas on what should we type into the AI for the video!"
Dude if your kids are still children this is happening before they hit adulthood
I think we treated Image Movers Digital too harshly. At least they know how to animate realistic remand better than AI, and they don’t look horrifying.
Eating spaghetti through your noise is scary
3:36 "Texture Sticking" how do i Searchterm the meaning of that? It doesn't know that I want the meaning of the whole phrase.
Books to film yay!
does the text to video have censoship?
You have to pay for the XXX version 😉
i love how ur thinking character has 6 fingers like it was AI generated
now this video is 10months ago, and we have sora💀
These are deeper and scarier than the uncanny valley.
Song?
Ok the Darth Vader one was pretty funny. 😂😂😂
I am a goofball, I love the flicker
Where can I make my own?
Haha I've made an entire fake trailer using this
Nice
Teslas FSD, and camera system has integrated hardware and software that monitors and generates temporal coherent modeling and models. Musks new ai system is going to be a beast.
talk about midjourney stoping thier free trial
and will it return
I Want an Ai That can upload and absorb the skills easily to learn anything faster into my mind. Imagine, learning to play a Piano you can master it for just 4 days
there is time for that level of technology, but we already have internet that allows you to learn what you want for free xd
What you are looking for is a ""Brain to Computer Interface, BCI)
The company Neuralink is developing that right now. A company from Elon Musk. They even have a CZcams Channel.
You can search for videos with valve founder gabe newel who talks about the possibilites with Brain Computer Interfaces in gaming.
To experience a game as reality.
Would it be worth doing it ? Because if u could do it everyone can , i mean u can just listen to a song , and it would be just as rewarding
@@throttle4418 People like me oculd care less about the way we just want the ends
Is it an app?
ty
I don't even want to compete with people, but let me live my fever dreams, they me what website or program I need to buy.... someone help me please.
we should've sticked with the images
THIS VIDEO GOOD
Soon we will have text to movie...
Papa cloud
I've got an AI femme bot from 2038 invading my channel with some retrofuturism weirdness 😱🤯😵💫
What's the name of the AI
Ok AI is getting out of hand now
Imagine what hold for us in the future
1:59 hahaha xd;
0:04 Elon Musk literally snorting spaghetti lol
No way these jokes are GPT generated, if it insulted obese people lie this there would be an outrage
1:49😂
Why celebrities and musicians are scared to death and I don't care! Lmao
Oohhh God!!! This is gonna he dangerous. Now I see why you don’t show your face. They’ll A.i you too.
noicrle
Bro what did I say here?
also these damn bots randomly liking comments
ai cursed
it's inevitable
Not cursed
Im surprised chineses corporations are making anything open source.
This is gonna be huge on onlyfans
Ummm why don’t they use 3d assets and train them with ai to move their bones correctly and then later overlay the realistic generated image