Create Videos Instantly with Meta's Text to Video AI! - Text to Video Generator
Vložit
- čas přidán 28. 09. 2022
- Meta Make a video text to video AI - makeavideo.studio/
Make a Video waitlist: docs.google.com/forms/d/e/1FA...
Emad's Tweet: / 1575499753445789697
Cogvideo: github.com/THUDM/CogVideo
General AI Playlist: • General MattVidPro AI ...
MattVidPro Discord Server: / discord
My Twitter: / mattvidpro
The AI I Use to EDIT VIDEOS! www.descript.com/?lmref=nA4fDg
Second Channel: / @matt_pie
Helpful Minecraft Tutorial Playlists:
Installation Tutorials for Minecraft: • Install Tutorials for ... Minecraft in-game Tutorials: • Minecraft In-Game Tuto...
THE THUMBNAIL USED IN THIS VIDEO INVOLVES THE USE OF AI-GENERATED IMAGERY
Thanks for watching MattVideoProductions! I make all sorts of videos here on CZcams! Technology, Tutorials, and Reviews! Enjoy Your stay here on the MattVideo Productions Channel, and subscribe!
Leave Suggestions, Thoughts or a Nice Comment! Bye! - Věda a technologie
The craziest thing about this is the fact that we are still in the early stages of text to video ai. If this is what we have now imagine what we'll have a year or two down the line! Absolutely amazing
What a time to be alive!
Yeah, we are even at the early stages of text to image! This is absolutely insane! And it will only get better from here, imagine just a year from now! Two more papers down the line! What a time to be alive! 😊😊
I love how artists attack these websites trying to prove their art is being stolen. When a majority of these artists provided their art contractually knowing what would happen.
Give it a month!
It's insane how quickly everything is moving.
It's not gonna be very long until we hear it again...
"Dear fellow scholars, this is Two Minute Papers with Dr. Karoly Zsolnai-Feher."
lol, yeah he's probably making the video as we speak 😂😂
@@pon1 yeah
Wwwwhat a time to be aliiiive!!
This is jaw dropping. A year ago we didn't have text to image yet. Eventually you'll be able to enter text and get out a custom feature length blockbuster movie
@Nicholas H Yes... on the other hand, excel didn't kill the accounting profession. Maybe animator's will just work 1/10th as hard.
@@energeticwhirlpool8666 Imagine AI code generators
The acceleration of progress continues to amaze me. DALL-E 2 was only released 5 months ago... *five* months. And it was the first well-known text-to-image AI (DALL-E 1 had a little hype, but it had enough issues to be less than amazing to most people, and wasn't public). FIVE. MONTHS. And now we've got companies competing and producing imperfect, but highly coherent text-to-video AIs.
...what is time? Remember when we used to consider AI progress to be on timescales of "5 to 10 years"? We've cut that by a factor of 12. Jeebus.
Transformers: brain-bots in disguise 😁
Timescales was wrong. Really this models have a problems with small details.
@@Nandarion yeah it sometimes makes weird striations when it shows a city of buildings in the distant background.
I can see inputting an old photo from a family member on holiday and turning it into a seemingly alive and logically animated environment as if it was a recording from back then.
@@Nandarion ? I didn't say the models were perfect, I was commenting on the rate of progress from one model to the next.
in future we will propably have au generated video games
The pace of improvement is so astonishing it's not an overstatement to call this metaphoric modern magic!
its absolutely true what you're saying. I thought about a year before we'd get anything like this and I guessed the processing would be astonishing. Really only the biggest companies with massive server farms could pull it off. I'm trying to just accept this is the pace of change now, just enjoy the ride of what's coming week after week haha
@@ClayMann I expect very soon, through VR/XR, we will experience a customizable "space" in which we can talk to the ultimate AI accessing anything we choose to explore.
@@Zebred2001 that's an interesting way to put it. When I first saw Dalle and there was talk that this could potentially be done with video. That really got me thinking that if an A.I could learn the tropes and a million other things about TV and movies, it could create completely new ones based on your suggestions. Another season of Breaking Bad. what would that look like? Firefly Season 2, the show that got cancelled before it ever really got going. Or completely remaking movies in a new way. Like Marvel What Ifs..... It seemed so silly to imagine that but then Dalle seemed equally impossible a year before. And just how with models like Dalle you can mash up genres. Breaking Bad mixed with Star Wars. You could do the same for movies. What if Star Trek and Star Wars collided?
I do think for a long time to come the processing for longer videos will still be an enormous task so I can't see real-time imagining in a photorealistic way happening...well ever as far as we are with technology today. Maybe in the future, lets say 10 years from now when processing is a million times faster than it is today. Exponentials really punch up to big numbers in no time at all.
I think really what you're describing is the holodeck for Star Trek and I don't think that idea is lost on these researchers working in VR and A.I. I think that's the top of the mountain everyone is climbing towards.
Hey Vince Klortho Do you know Key and peele by any chance?
@@lukelouis4080 I know of them. Why do you ask?
Well, we’ve seen text to text (GPT3) we’ve seen text to image, we’ve now seen text to video, get ready for text to video game.
Edit: Wow #1 comment.
Already on its way with codex!
@@MattVidPro Oh god how did I not know about this. This is incredible.
Crazy that we have "text to virtually anything" rnm
Text to music
Text to hyper realistic universe simulation
I imagine we are on the verge of open-source text-to-video generation to the extent that you will be able to verbally describe feature length movies into creation.
You could input text from a book and it will make a movie out of it.
@@piotrszmyt106 I'd like to see how that compares to someone's mind's eye of what they are reading.
@@piotrszmyt106 man can't wait to generate a movie out of *never gonna give you up* lyrics
As long as these come out open sourced, the speed of progression will multiply exponentially
I think we will see video-latent-duffusion
I think that the knowledge that's driving these things is becoming well-known by a lot of people. Even if some companies try to keep it away from the public I don't think they're going to be able to.
I have a feeling it was released specifically because of the explosion of success and popularity mid journey and dall E have had and then seeing the customised open source stable diffusion applications.
To put it simply, meta going open source boosts their productivity tenfold overnight.
@@MonsterJuiced but then they would be losing money.
@@dibbidydoo4318 that's not how money works
Fucking crazy, even i am shocked, even with that i knew DALL-E back in 2020 when almost nobody knew about its existence with seeing all that evolution with text to image AI
I learned dalle's existence September 2021
The coherency of all the individual hairs and other small details like that following steadily throughout the frames is pretty astounding.
Nice shoutout to The Dor Brothers! Good to see him mentioned here since the two of you are who I get my AI news from 🤓
His channel is great!
At this point by the end of the year we'll have aovie of human history made with ai screening in theaters
I am so hyped for this! What do you guys think?
I think its the greatest thing ever
Just add dreambooth to this, and it will be the best artist AI ever
I don't care how spectacular this META thing is. Anything tied to Zuckerberg is a no go for me. He completely lost me as a user of his products when he put profits above doing the right thing with his platform. And he set up his platform to make it as difficult as possible to move your information to other platforms. I understand this from a profit motive move, but I believe you should win people over on how great your platform is. You shouldn't force people to stay by making it difficult for them to move to something they perceive as better.
Amazing. Thanks for covering this. You're videos have been really great to keep up to date with AI text image tech.
-Ken
@@henrythegreatamerican8136 who cares though Google does the same thing. But you still use their services.
The absolute hateboner for Facebook is a bit much imo.
Holy shit...(imagine that being AI animated lol)
that's beyond crazy at this point, next thing it would even be able to create a corresponding sound effects based on the given video...so insane.
I'm pretty sure there's already an AI that does that for silent videos.
@@WhiskeyBlack777 Haha so they would only have to integrate it with their AI image gen, can we imagine how far they will take this tech in like 3-5 years from now? wow...we live in the future, it is scarily advanced.
@@21EC imagine AI code generators
@@ViralVibesReal Yeah I already thought about that, like what if you give a neural netword AI to program stuff and create full blown apps or games, it has no limits to what it can achieve and do.
@@ViralVibesReal BTW...4 months later... (more like 3) and we have now ChatGPT...which can totally create and modify scripts and codes of programs and what not..crazy times.
Is funny how human artists also struggle when drawing hands, AI: “Right? Hands are actually hard” 😅
Now THIS is something to invest in, Meta.
imagine the horrors you could make with a more advanced, uncensored version
R34 content will be off the charts.
this is already a horror show . this must stop .
@@samankucher5117 that's the thing, it won't stop
"Man being shot"
we'll have AI movies on prompt in just a few months, the meme potential is amazing
Actual quite a bit deeper, even than that.
The implications of fully realized, frictionless imagination is going to make the industrial revolution look like a quaint park visit.
And, it is only one of many such things on tap, hurtling towards reality, soon.
I believe games are going to be next. For instance I want the D&D rules for a space adventure in the theme of Grand Theft Auto and the AI will make it that game.
Game development is far more advanced than anything that ai is remotely capable of doing some games can't even run properly on current gen graphics card without sacrifice
@@samuelkibunda6960 It'll exist this century.
@@orlock20 yeah definitly but im calling it cap if itll be in my lifetime but itll be possible to generate ps1 or 2 styled games before 2050 the biggest issue with video game generation is that there'shardly any great or ground breaking open source games to use as a data set
The AI image to image animation is exciting - I've been waiting for AI to make independent animation easier and more viable since AI images started making their strides, I just didn't think it'd happen so quickly.
Nice, lets hope more companies gonna make open source variation
It is so amazing to see how quickly these AI tools are developing! So looking forward to a Netflix AI which shows me just generated movies and learns what I like and what not.
I don't know if that's good or bad but I'm excited to see of how this field evolves
@Nicholas H Society is a tremendous force. What is and isn't accepted or hoped-for will be decided by the way the world responds.
@Nicholas H With all due respect, good luck making that happen. Artists will need to find their place again, much like with the invention of photography.
Rejecting such a huge leap in technology is futile. At this point, it's inevitable that the landscape of art will dramatically change. We just don't know how it'll turn out.
@@creativebeetle So sad we cant have good tech without it replacing people and causing loss of jobs, i wish we lived in a better society but the qorld is not fair!
@@samuelkibunda6960 I sympathise, man. I'm an artist, myself. But hey, things come and go in equal measure, and revolutions in technology tend to open unexpected possibilities.
We are notoriously bad at making predictions of future technology, and even worse at predicting the resulting changes in art and culture. At heart, artists are simply people who thrive on creativity, and I think it's unlikely they'll surrender that passion in the wake of a new and uncertain age.
This is extremely amazing. Thanks for this AI text-to-video tips.
Just imagine, in 10 years you could write a novel, and AI makes a movie about it, with the actors you want, or even you and your love as main actors. Or you write a fairy tail for your children, and the Ai makes a movie for your children out of it.
That sounds incredibly depressing. "Yay, I can have whatever I want without relying on other humans in society!". Everyone just sitting in their own content consuming world asking an AI for what they want, instead of collaborating with other humans to create these things. The advancement of this tech potentially replacing human's need for creativity will be a massive trigger to the downfall of humanity.
@@invertexyz Actually there is a science fiction novel by Herbert W. Franke , called ''The Orchid Cage'' from 1961 (!), where a group of people finds an abandoned planet, with signs of a highly advanced civilization, which doesnt seem to exist anymore. That group of people on the planet use the planet as their ''playground'' (like an egoshooter), not caring about the lost civilization on it. But one of the group, the protagonist, begins to wonder ''why did that civilization on that planet go extinct, when it was that high advanced ?'' ... He forgets his ''egoshooter-game'' with his friends, and tries to figure out, what the reason was.. He finds libraries of that civilization (the books there are projecting their content direct into his brain) - so he sees: that civilization had quiet the same development as that from the mankind.. So he still doesnt finds the answer : ''why did that civilization go extinct ?'' ... Later on, he gets captured by a ''guardian robot'' , which reveals him the answer... that civilization on the planet in fact never went to extinct... it just ''transformed'' more and more into beings which are more and more connected with computers. The protagonist founds those people underground in glas-tubes, where almost only the brain is there, and few muscle-tissues and skin-tissues hanging on it.. and those brains are connected with an AI ... The protagonist wonders, because in the library books that people once looked like humans. And he asks the guardian robot : ''they dont have feet, arms, eyes anymore''... The robot answers ''for what ?'' ... The protagonists further asks : ''they even dont speak ?''... and the robot again ''they dont need to speak !'' ... Turns out: all those beings get everything they wish, right into their brains from the AI .. those ''people'' didnt need to move, or talk, or anything. They ''live'' the best live they can imagine. The computer gives them everything they can dream of , as a indistinguishable ''reality'' right into their brains.
Remember the author wrote this 1961 , were there was not even imagined such technology like ''Virtual Reality Headsets''
Turns out later, that the protagonist itself didnt ''traveled'' to that planet, but ''beamed'' to that planet by a computer on earth. That ''person'' he was on that alien planet was only materialized as ''strong man'' from the computer, while the real protagonist was on earth - and only connected by ''brain-waves'' with that ''materialized strong man'' on the alien planet. When he disconnects with that ''alter-ego'' , he sees himself as very weak person, who spend almost all of his life in his computer-chair, surrounded by all of the computers which are doing everything for him... they cook for him, they feed him, give him pleasures, everything. He now recognizes: humanity is very much going the same path as the civilization on that alien planet... being just a consumer - and doing nothing on their own... degenerating to a wobbly brain in a glas-tube , connected with wires, and living eternity as passive consumer of computer-technology.
The protagonist is scared about that .. and he itself can barely move, because he barely moved in his life. He tries to find rods on which he can stem his bodyweight... Than he tries to destroy the computers in his apartment.. as much as possible.
And then he goes out of his apartment,.. barely on his feet because he is so weak... and sees the sun for the first time - after decades in his apartment. The streets in the city are empty (because no one needs to go out, and everyone stays in his apartment, where the robots and computers doing everything for him)... The protagonist on the other hand decides to try again: live a life as human being.
I can totally see that we are going in that direction. We have robots now who washs our dishes.. washs our clothes (washmachine), mowing our lawn, vacuum cleans our rooms. We need less and less to go shopping .. internet shopping makes it that the things we want come right to our doorstep. And we have virtual reality, and the screen , and every movie, music etc on our fingertips. We dont need to go into the theater.. because we have almost everything at home. We even doesnt need to meet people, because there is telephone, skype, chatrooms, etc ... and even self driving cars begin to evolve and serve us more and more... Hell we dont even need to stand up anymore and turn a switch to turn on the light or off. We have ''smart homes'' , where we can say ''Alexa light on'' , and it goes on.
Sure we are still far away from the ''orchid cage'' ... but i can see it absolutely in the not so far future. In that sense: the future is bright and dark at the same time. I will get indistinguishable who is the master, and who the servant, if we do anything to depend on computers and technology, so that we can enjoy their comfort. The machines doesnt need to enslave us, when we are very willing to depend on technologies to comfort us. The question rises: what is a human, and how do we want to define what a human is, in the near future.
And to imagine: Franke wrote it 1961 - more than 60 years ago (!) , when personal computers didnt even exist , is just mind blowing.
@@PygmalionFaciebat Thank you very much for sharing, I always had some similar ideas to that story!
@@PygmalionFaciebat
Yeah thanks for sharing. Those early sci-fi writers were really prescient.
Little known fact you may find interesting;
the 1st patent for VR was in 1962.
@@bellezavudd Yes i know about it... It was called Sensorama
Absolutely mind blowing. Makes you think if this is already possible now, wether youre gonna be able to trust any videos in the future.
I'm guessing that, in the future, more and more videos (and images) will need to be digitally signed in some way so that we can verify the source and authenticity.
After Facebook's bullshit with forced ads and locking Oculus users to facebook accounts, after they said they wouldn't, I'm glad StabilityAI stuck it to them with that tweet. Hopefully they can deliver. The last thing I want to do is touch anything by Facebook and all the strings that go with attaching yourself to that platform.
It’s about to get spooky!
Best part of this is they released the paper, so hopefully it turns up outside of the Meta-verse. Made me feel grubby even going to look on their ai site.
This seems like a great way to completely kill off the stock footage industry. So there's that =)
Jaw dropping. Thank you, CZcams algorithm, for bringing me to this video. A user of Stable Diffusion and now also DALL·E 2 (since yesterday) here.
AI will change our world in ways that maybe we can't even imagine, and it could go so wrong.
welp, here we are... 2 papers down the line... already
I love the sliding dude on the left side of the dancing robot animation
Looking at the couple walking in the rain... they do appear to have the same legs. Not sure what to make of that.
This is great. Like… i have no words
this is the first of a brand new type of video creation. It isn't animation, which is why it doesn't have that "weightless" look that CGI has often. It' isn't practical, so it doesn't have the jankiness of animatronics or in-camera trickery. This may be the next step in non-real film.
Continuation of Zuck's life works. Taking something like an open source chat program (originally used so he and his collage buddies could fat shame campus girls ) and everything else open source ( free) that is already out there and claiming it as if he invented it. The Metaverse existed in the 90's, and 'Meta's ' version is not much better than what existed back then (also open source). Many other virtual world options are far ahead. The only thing that makes this Meta's AI is because they copied it off git hub where anyone can download it, including the new ability to animate your diffusions which came out days ago. (open source). And yes the zooming traveling camera is part of that release. Not to take away from the tech, it certainly IS impressive. I have several versions. The level of advancement and resolution is all parameters and prompts. As long as you have a powerful enough system, you can already do it too on your local computer, if not you can rent cpu power on google's colab. You didn't need your Mom's Facebook chat program. You don't need your Grandpa's 3D goggles virtual world chat program. You don't need Meta.
this is creepy. the fact the AI is making images and videos using stolen art is unforgivable.
Yeah! I've never seen anyone go to Art Station, Pinterest, Instagram, ect. And create a PureRef board of inspiration and transform those images into a new one.
We artists all just wear blinders 24/7 and are oracles of original art!
Hyperbole aside I don't think that art created to be new transformative art with very few changes made to the original is acceptable - when Bungie's Halo was given to 343 we didn't necessarily expect for 343 to change everything "because while this is our OC now we didn't draw him this way so it's stolen".
I think that because we can't seem to provide source images / "inspiration" from the Ai that this is an unfortunate reality that sometimes the generated art isn't going to be very transformative. I highly recommend treating all ai generated art as if you found it on Pinterest. Do take inspiration, do not steal.
@@OverbiteGames
do not compare a human's representation of reality and his imagination to the photo bashing done by a faceless lifeless lines of code if that is how you live then don't talk to men or women again when you such for a partner just used a dildo or a pocket 😻 or better a sexdoll it is easier to get and it's cheep and it has no opinion just like you it will say anything that is told this comparison (AI is just like humans there is no art theft) ive heard it a million time it's a copy paste argument and it's a foolish one at that .
will if a human is just a photo bashing algorithm like you said i would like you to go and looks and a bunch of art on Instagram and draw something as good as that for me if it is something that easy.
the company used the stolen art not the ai
@@frost_spirit3314 true Steven Zapata maide a interesting video about that the corporations that stole the art was pretending to be a nonprofit organization to take that art and use it and long and behold they are not a non profit organization.
I don't understand how fast these ai advances. It's super overwhelming!
Perhaps they have a secret deep learning AI which can develop other deep AI solutions...? a crazy speculation lol
My mind is officially blown!!! Tell me if I'm Dreaming, I'm not sure.
Next year text to video generated videos will become popular on youtube.
Imagine walking around in a metaverse generated by A.I.
In a few years that's probably going to be pretty cool. But there will be a period of horrifying experiences.
Plot twist: you are already walking around in a universe generated by AI
@@throwawayidiot6451
Actual plot twist :
you are walking around in an experiencial world generated by your nervous system.
I am very much impressed
i really apreciate your help with dowloanding this software
Videos are simply sequential images with slight changes, it's literally stop motion animation, of course this was next
The pair of images option for image imput is huge for animation to make the in betweens
Didn’t even think about that!
Nice video, little buddy. Can you show some demonstrations, buddy?
So, simulation theory confirmed?
Nah but we are simulations on the brains of ape's.
Not yet, maybe in 10 years or less
Super cool next generation AI continues to excel.
Everyday I wake up to new stuff
Combine this with a recommendation algorithm and you can have CZcams without CZcamsrs, but only AI.
that's a nightmare.
Matt - Is it possible to make a 3-d video of which to use in our school science project? We’ve created a homemade hologram projector which we learned from various youtube videos.
Try metahuman and character creator and that could be ai as cgi people to text to videos and images and blender also?
Ok meta, you have my attention now
I think the GPU requirement would be really large for this, it could be around a dozen times an art AI to run this locally and would take an hour to generate a video in high end computers.
my only answer to that statement:
high-er end computers
We're truly in the endgame now
That teddybear with human hand is actually terrifying
Can't wait to get my hands on this in a few years, feed it an image of a turkey sandwich, and completely forgot about it afterwards.
Translators, voice actors, narrators, artists, programmers now content/potentially movie creators. Any other professions who have people left for ai to make nervous? lol
We're about to turn into those floating people from wall-e.
Definitely interested.
AI to music, article writing, customer service agent (that you think is a human), games. Full novels? Long films? CZcams channel?
Are ther any style transfer to video tools available?
This is a wild trip
Hi there! Does anyone know how long it takes to render such a sequence and on which hardware?
Nice!
Nice video. Little tip, you don't pan up, you tilt up :)
horrifying that someday people could probably put in an image of your face and a text prompt and make you do whatever they want you to do. someday nothing will be believable.
AWESOME!!!
Creating your own photoreal universe in VR..............F ! This could be the Metaverse we want
If you guys want to be a youtuber better start now before it gets automated!
Fr tho gonna make a 15 min youtube video in 5 mins in the future
well we did it guys
the next will be text to interactive simulation
Phenomenal! Scary!
Is this available to the public yet? Which Github?
animators are definitely next on the unemployment chopping block
What a time to start a CZcams channel!
Instant GIFs.
creeps me out for some reason
it should basically the AI stole art and videos from artists and random people and photo bashed it to make something to replace actual human artists .
@@samankucher5117 stop spreading misinformation
here are some links to videos that actually explain how this tech works
czcams.com/video/0QczhVg5HaI/video.html
czcams.com/video/hfMk-kjRv4c/video.html
czcams.com/video/gvNdCRe3T-g/video.html
Wow, this is very exciting indeed. With "Pair of Images" - I wonder if you could get it to produce a seamless video loop by giving it the same image twice.
There are already seamless videos made from Text-To-Image AIs czcams.com/video/0fDJXmqdN-A/video.html But I bet a Text-To-Video AI would be able to create this much faster and easier
O M G
And now we have sora, we've come so far in a year and a half
Can you imagine? In _at most_ a couple decades you'll be asking your TV not to fetch a movie, but to make one. Probably better looking than modern Hollywood, and certainly better written!
And it will save ur generated show so u can make variation episodes endless episodes and then it can probably generate spin off series's
I'd prefer to watch what others created. Sure it'll be cool a few times to experiment with the tech, but never knowing what's happening and being able to semi-rely that it's good via reviews and whatnot improve the watching experience.
An example is Novel AI. Basically writes a book for you in a similar way you mentioned, yet I'm still going back to books written by others. Even AI generated by others is better.
or just generate based on your favorite fiction.
@@caenir that's true, but what if you want something specific? Maybe a funny plot idea, maybe a fantasy, maybe just your personal story. The possibility for data mining are spooky.
@@4.0.4 Sure, but I see value the value of that being sharing it with your friends. For example, when I was in school we used to write hunger games parodies of our classmates. Then a website came out that did that (saying who was going to be eliminated and how). But that was fun because we'd show each other.
I've played around with novel AI a bit, and I can't imagine writing a novel in it to enjoy just for myself
Thanks.
You're welcome
PANDORA'S BOX HAS OPENED!
Cool
Bonkers
Thank god, we'll be able to feed the Silmarillion to the AI, and completely forget Rings of Power ever happened. 😅
Or even the entirety of the old star wars eu, time for the darth bane movie trilogy.
Googles looks like a more refined for videos models of the SD
And I though text2image and text2video were already moving at breakneck speed for the past couple months.
It feels like everything is accelerating week by week, day by day
and text to 3d model
Have you seen DreamFusion text to 3d? It's incredible, this is advancing too fast
Ive seen snippets of this! I cannot wait until this market matures more
oh no. The legs of those unicorns.
already signed up for it
remember how speech generation evolved? interesting relationship it had to speech recognition. interesting how it happened in reverse this time... isn't it!?
Cool and everything but Meta history can only mean one thing, "ALL YOUR VIDEOS BELONGS TO US!" 🦎face
The endgame metaverse is text to video
coool
if someone were to create a dataset of anime and train it on this dataset, it could maybe be able to do inbetween frames in the future
Wow, in short time people will make their own virtual reality worlds.
I have a *MIGHTY NEED!*
Most new artists would agree that hands are fr@cking hard.
damn, meta is moving at some weird place
woah