Google Casually Drops the Best AI Video Generator We've Ever Seen
Vložit
- čas přidán 15. 06. 2024
- Lumiere is a text-to-video AI model from Google that uses space and time to create realistic videos. It's a time-space diffusion model that transforms images and text into AI-generated videos. Lumiere can create video clips up to five seconds long and can animate still images.
▼ Link(s) From Today’s Video:
Google Lumiere: lumiere-video.github.io/#sect...
► MattVidPro Discord: / discord
► Follow Me on Twitter: / mattvidpro
-------------------------------------------------
▼ Extra Links of Interest:
✩ AI LINKS MASTER LIST: www.futurepedia.io/
✩ General AI Playlist: • General MattVidPro AI ...
✩ AI I use to edit videos: www.descript.com/?lmref=nA4fDg
✩ Instagram: mattvidpro
✩ Tiktok: tiktok.com/@mattvidpro
✩ Second Channel: / @matt_pie
-------------------------------------------------
Thanks for watching Matt Video Productions! I make all sorts of videos here on CZcams! Technology, Tutorials, and Reviews! Enjoy Your stay here, and subscribe!
All Suggestions, Thoughts And Comments Are Greatly Appreciated… Because I Actually Read Them.
-------------------------------------------------
► Business Contact: MattVidProSecond@gmail.com - Věda a technologie
Double Upload Today!!! Sorry I just HAD to talk about this, it blew my frickin mind!
This is wild…
I'm glad you did
@@The_Questionaut appreciate!
Now another video of you'res to watch today! Already a great day.
wow really Omer Bar-Tal and hilal and others they are incredible, what an incredible data ai talent.
2 years ago, we couldn’t even make a photo that wasn’t in the uncanny valley, now we even have videos that can escape it
That's true.
Imagine what will be possible just two more papers down the line...
@@denisgabriel4645 @TwoMinutePapers is a great channel :)
That's exponential growth for you. And to be quite honest with you, I'm pretty scared. And I'm usually not easily scared...
@@idib1739 don’t worry ai companies have already started training their AI’s on morality and it likely will become even more trained in the future
@@denisgabriel4645 Only took 3 weeks lmao
Google is the friend who invites you over to play, opens the toy chest to show you all the amazing toys. You: "Cool, can we play with them now?" Google: "No, these are mine!" Slams chest shut
"Play with them? Why would you want to do that?"
"...because they look fun?"
"Hm... in that case, let me put all the dolls in my dollhouse and set it up the way *I* like it, and then you can play with it that way!"
"But... can't I just play with the toys the way I like playing?"
"NO!"
It's weird how Google's approach to AI is like Apple's approach to operating systems...
LMFAO
Like the rich kid with the Nintendo in the movie 8-bit Christmas
Fact 😔
Playing with toys is inappropriate and I am unable to help you with that request. Do you have any other incredibly dull and boring questions I might bother with?
Finally a way to make low-budget social media video ads, once this gets released! I’m a marketer, and all the ideas for video campaigns go out the windows due to budget. This could be it!! 😁
The first thing I saw was the motion.
this model had the best looking motion I have ever seen out of a video generator!
AI moves at Hyperspeed this year will be mind-blowing!
Yep. Mass business adoption is right around the corner. These AI art tools are not even where the big bucks are. Imagine what they will release where 100B+ dollar industries have an interest.
I disagree. When I first heard about it, I assumed it would be light-years better than it currently is.
And now, I'm getting bored waiting for it to catch up to my initial, modest expectations. Maybe in 10, 15 years...
@@SpaceCattttt what are your expectations? We have made very big leaps in AI in recent times. It seems to me like you might have slightly unrealistic expectations
@@SpaceCattttt Judging by your "modest expectations," I bet that you're one of those people who are disappointed because we won't have AGI in 2024.
@@denisgabriel4645 The sooner the better, yes.
I find it really annoying that Google always comes out with a paper for a small group of researchers with PhDs and no real software for regular users. For good PR, they should release a working software immediately and make it available on Google Cloud, even giving some free credits for testing. Otherwise, don’t publish the paper, damn it.
Matt shouldn't get us hyped for it within the main title if it's not available yet but still should spread the news about it within the video
Teasing you.
I find it annoying that I’m too stupid to read the paper and duplicate the results for myself. Ugh google should hurry up and give me the ability to read and synthesize data.
Spot on
Google always doing this ans annoying. Good thing another dev use this as badis and create fasterfor end users.
Three weeks ago:
Google Casually Drops the Best AI Video Generator We've Ever Seen
Five days ago:
Open AI Releases the BEST AI Video Generator by FAR. Sora Text to Video
This does look next-gen, I hope the actual product doesn't disappoint when it releases. I recently just finished creating an AI video using Pika (not published yet) in anticipation of Chinese New Year, and I can't imagine how much better it would've looked if it's made with this.
I wanna be able to test these myself before I get too excited about their models. These guys and other folks in general to Cherry pick their best results and not buying it until I see it.
What I really like about this is that (even if Goggle doesn't release it) this thing will create much more competition, which will result in even faster AI video development.
this is a good point !
When will Google release it?
I predict you'll do DomoAI next (sadly a discord video to video generator).
they reached out to me to do sponsored content its on my list
I can't wait to animate some of my book covers for promos! Thanks for the heads up.
Imagine when these generations can be done in real time, imagine the impact it will have on videogames
More like the movie/entertainment industry
@puffytheangel483 It'll destroy the movie/entertainment industry lol. Any dude who knows what AI is could make a Marvel tier masterpiece on their own.
Videogames still need to be done in real time and be controlled by a player, so AI would give practically unlimited graphics
@@xitcix8360 I also agree that AI video generators destroy movie industry. All will be artificial generated, poor quality and show world that dont exists. Most of peopole wants to see real world and immerse to movie scenario.
@@xitcix8360 yeah this was the point I was insinuating per my previous statement
@@puffytheangel483 True
The demo video is not what people are going to be generating when this releases
Might be better, might be worse. Depends on how much they improve it from now to the time of release.
@@jacobnunya808that’s not what he meant mate
@@hrm6750 ahhh I see. Depends how good it does with people I suppose. Worth a shot I guess.
@@jacobnunya808 In the miraculous event they release the model open-source... there will be fine-tuning 😂
@@IceMetalPunk it's really just a matter of time before an open-source version exists.
I’ll believe it when I can use it.
Hope they put this in Gemini Ultra or something so we can use it
Me too!! But I doubt that it would be anytime soon TBH
Matt does it again with another great overview of new AI tech. Always so clearly explained.
Untill it's actually released, it's vapoware like the two previous Google AI video generators. There is a good chance that this video will not age well ;)
By the time they release that thing, Pika and Runway will have something way better.
@@denisgabriel4645 Soral is already better it seems
8:15 "Dog riding a snowboard" - explains the CZcams Amazon UK ad?
we can't take their promo video at face value
Google finally getting ahead in the a.i. space. Their image generator looked so far behind, but it looks like they have something that they can take the lead in.
EOY at least one quality 30min fully generated short film
At this point, whenever I hear "Google AI research", my response is: 🤷♂ It looks nice, but meh. Unless and until we can actually use it, I believe it only 70% and care only 10%.
I'm still looking forward to video generating via pre-written screenplays, in which there are consistent generated characters who say their lines from the script, and move in the environment created from the script's requirements...
When it comes to still photographs, Ken burns, ate your heart out.
Hey Matt, saw this 11:20 and really getting Hitchhiker's Guide to the Galaxy Vibe. This should go into the Rabbit R1, what do you think
Will be interesting to see open ai/microsoft's answer to the video scene
I’ve only watch a little bit here, but my first impressions is that we’ve seen in the past where Google is willing to use misleading tactics to make themselves look further ahead than they actually are. I’ll believe their hype only when other non-Google people have access to their tech and consistently get the same good results.
WOW.. that and the new Google Labs stuff is really cool.
Already seems like ancient tech compared to OpenAI's Sota
hope we will get the code and model open source, very impressive
The last time google dropped a trailer it was fake
I go over this in the video! I agree skepticism is good!
It is a phantasy, it is a commercial showing how they imagine it will look like in the future if they ever are able to create it...+--
you mean aspirational :) Cool. @@a64738
RIP Pika, all of the videos on my CZcams channel are images generated in Dalle and then animated with Pika Labs, not anymore!
Pika and any others will be fine. Google don't release their toys so everything they claim is meaningless.
Great video (again). Thank you for doing this 🙂
I doesn't wipe the floor, until it's released IF EVER
The thing with Google is that by the time they do, if they ever do, release this, Pika and Runway will have caught up or exceeded the quality, whereas this will stay the quality we see here, making it old news.
O:50 this is the most impressive bit for me, taking an image and animating it to a prompt at that quality.
Google plays a bit fast and loose with their demos tho.
I wonder if they used an integration of LLM interpretation to understand the context of what it sees in the image, like: "If camera moves and the sun is peaking through the leaves it should fade out when leaf moves over the sun"
Very fascinating though, great video!
Amazing! It will be fun to play with when it comes out! If any developers read this, I can't wait for someone to generate an AI that removes green screen backgrounds from raw video. Right now, it is still somewhat of a job in Final Cut Pro :)
The cornerstone of all human experience has been 'seeing is believing'. Now with Generative AI that's about to change to 'believe nothing you see'. In 10 years things will be very different. A profound personal and social transition for all of humanity is at hand.
That's not really true. People used to believe in gods and even sometimes give them human sacrifices just for a chance that they will bring them rain or something. It's not really "seeing is believing", it's more like "hearing an authoritative guy say stuff is believing".
Bruh i dont even know what we are gonna have by the end of this year
The bear dancing though lol
lmao ikr
In the drifting Lambo segment there is writing on the road going behind the lambo but not at the front of the lambo.
WOW! ....I'm absolutely blown away by this.
Things are changing for film in every way imaginable.
Crazy how real it looks! I wonder how long before we can't tell the difference between AI and humans.
Great video, but when will it be available to the public?
I love that we've reached an era where someone says "Check out this awesome AI generated video!" and you're like. "That's just a normal video, dude. " You have to take their word for it. I guess the smart thing to do is to generate hyper-real videos of things that just plain can't or don't exist. Which is more fun anyway. :)
"When I see it in production, I will believe it, because much marketing is only half true."
it's ok. Can I use this Lumiere our of the box, GPT style? Nope. Same approach as gemini pro demo.
How to use this Ai and where to find it's website?
Very good. Nice features.
But for Google, this looks like an alpha test.
They don't really attempt any realistic full-limb movements over realistic backgrounds.
How you can believe google after the recent event
Can you explain lol
Google still didn't finds out best AI for edge detection in portrait mode.
Fantastic to see what is possible. Of course, since this is from Google, it wont ever see the light of day though.
I think otherwise since they are advertising it so much!
@@MattVidPro I hope you're right, bro!
I still got samsung face generator in my bookmarks... they announced that 2 years ago😂😂😂
@@MattVidPro It's fair if you want to think otherwise but I'll believe it when it's actually available.
Super amazing. Hope we fet public access soon. Also why does he sound like a less peaks n valleys Kermit the frog, but in a good way?
Kewl subbed looking forward to more
I'm waiting for AI to get to the point where we can feed it pages of a Japanese manga and have it spit out anime complete with Japanese voice acting music soundtracks and sound effects in full motion animation.
In some way I kinda hope it won't happen (but it will)
Yes it would be great, but you know, I like to know that real people have actually worked on it, drawn, composed, and all that stuff and I don't want this industry to use AI to make things easier and cheaper
@@guerricreal people came up with math to make it possible. Being an animator is harsh, they should use all available tools to lessen their load.
@@guerric "I don't want this industry to use AI to make things easier and cheaper"
So you want things to remain difficult and expensive? 🤔 I'd rather live in a world where high-quality entertainment can be made by individuals instead of huge corporations.
@@Lar_me Well as I said I prefer enjoying art that was made by humans because I like to know that they've put their heart into it. You can see it differently but I do not
I'd say something like that is maybe 5 years away
It looks great!
Waiting for the cursed memes gifs to start coming from ai text to video.
Cool video! Luckily, I watched it entirely before making that 'those could be cherry-picked' comment. Still very hyped to test it, despite preferring open source, though.
Totally with you!!! Hopefully the research paper inspires open source developers
If the tech keeps evolving this fast or even faster, open source developers will keep so too.
Dope Content Matt 🤘🏼😎💯💧
By next year we wont need to step out of the house for a video shoot. No expensive cameras needed as well.
how do we access it isit free can we download it and do it offline ???
Looks amazing. But what is it going to cost?
Seems like the beginning of the video got muted
Matt you should make a video about robloxs Ai assistant tool its crazy haha
Google's demo is always better than the real product
Is it out yet?
looks awesome
Marines raising the flag background: "One of the most iconic photographs taken of the Pacific theatre during World War Two is the image that captured the raising of the flag at Iwo Jima. Taken by American photographer Joe Rosenthal on 23 February 1945, it won him a Pulitzer Prize.
The image depicts the moment that six marines hoisted a large American flag on Iwo Jima’s highest point. It was actually the second American flag to be raised on Mount Suribachi that day. But, unlike the first, could be seen by all of the men fighting on the island."
See film The Sands of Iwo Jima.
Very impressive. 👍
Sora has already gone further now and this wasn't that long ago
Clearly the famous painting look wonky.
be nice linking a Mic and as you talk it updates the prompt automatically
That is pretty insane!
It is a phantasy, it is a commercial showing how they imagine it will look like in the future if they ever are able to create it...
They didn't drop anything. I'll believe it when I use it
3 videos within 24 hours? yessir
Now let's hope they didn't stage this like they did with their ChatGPT "competitor"...
It is not real, it is a phantasy because it is a commercial showing how they imagine it will look like in the future if they ever are able to create it...... m,,
Mind = blown. The bar has been set higher gents
LTX studio looks awesome.....there's a waiting list. Movie makers first....then the rest of us in 100 years.
You did not mention how stable the image looks , the images are consistent characters , the clothes ,hair and faces doesn't keep changing . yeah if this cost less than a hundred a year for an ap , I'm in
I mentioned that, you must have missed it. It is temporally consistent
@@MattVidPro Yeah I did , Consistency that is the Holy Grail , AI video is useless to me without it , specially Image to video
I feel like some of these might be for pixel phones.
thanks
Unreal! Wow!
I remember the 60's as a child, watching Superhero cartoons. One of the main quotes from a Superhero who was fighting the bad guys was, "We can't let this technology fall into the wrong hands". Well, that is what happened. AI vids and other information are being developed by the bad guys, to be used IMO, to "distract us from reality". If we can't figure out reality,--------------------and brute can rule us. It is hard enough for the population to figure out reality today, as we live through the results of that ignorance.
"We can't let this technology fall into the wrong hands." That is what happened. Bill Gates started helping when they pinched him in 2000.
amazing stuff
A couple years ago text to video gave me nightmares. Now it is just boring, semi-realistic, non creepy videos.
Before you know it - indestinguishable
That's the true nightmare of AI video/ images/ etc. Completely derivative homogeneous regurgitated ubiquity. The creative equivalent of grey goo. Yum 😂
@@clintslowik6490 Maybe you're basing that off of the promotional materials, but put this in the hands of creative people trying to bring their visions to life, and these tools can be extremely powerful and creatively freeing. And there will be additional techniques developed to build creatively on top of these tools as well, so it won't be just the grey goo you are imagining.
It really fucks with my eyes somethings off about it. Almost like its blurry but its not
"as long as it's not available to the general public, it's meaningless."
I can really be happy universe ripped me out of creative business by health issues in this era where my abilities and talent are needed less and less ;) Makes me watch it ore fascinated than in fear.
If it’s actually released I’m not jazzed one of the most all powerful companies in the world would have the most video creating power over small genius ai based ones but it’s going to be what it’s going to be so will embrace the best possible for one and all.
GG, unless they picked through trillions of attempts.
I want to generate like this on my 10 series video card.
Matt for AI President.
Based on Google doctrine a little bit the Gemini results, I have to wonder this is not cherry picked or shown in such a way and just like the original Gemini stuff to where it's not an accurate reflection. It can do some of it but maybe not that level. I have serious questions because there are in an arena that has a lot of lot of competition and they lost a lot of respectability when they did that. I don't really kind of believe what they say anymore. I believe it when I see it. My own two eyes and nothing more. Their videos will be doctored. Maybe not complete lie but it's still a lie
1:58 no freaking way they misspelled "Origami" 🤭🤭
Did you just upload back to back? 😆
It's not released yet, right? Let's wait till we have it in hand...we don't want to get fooled again.
Google does a good job people walking when camera at the back.. other video ai failed
Probably takes an A100 video card with 80GB VRAM and a degree in AI software development to use it at this point, even if you did have access.
Thank you, Carl Sagan, Jr. 😋