Googles New Text To Video AI "VEO" Is Actually AMAZING! (Googles SORA KILLER!)
ฝัง
- เผยแพร่เมื่อ 2 มิ.ย. 2024
- Join My Private Community - / theaigrid
🐤 Follow Me on Twitter / theaigrid
🌐 Checkout My website - theaigrid.com/
Links From Todays Video:
deepmind.google/technologies/...
Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos.
Was there anything i missed?
(For Business Enquiries) contact@theaigrid.com
#LLM #Largelanguagemodel #chatgpt
#AI
#ArtificialIntelligence
#MachineLearning
#DeepLearning
#NeuralNetworks
#Robotics
#DataScience - วิทยาศาสตร์และเทคโนโลยี
All these unavailable demos...like idc until we can actually use it
Absolutely
True, GTP 4o, although better than 3.5 I some ways is not up to the mark with the demo vdo
Same!
Agreed. This seems to be what Google does a lot
Agreed. Alot of these AI news pages want to be like the Daily Blob with Eli.
maybe I should ignore the news for a day or two until they announce a 3D printer that spews out perfect humans.
Did you miss that demo? It was last week. :-)
Haha, yeah, every week there is a new leap forward
@@RetireandGonow the impressive thing will be ugly humans, just like we try to get ai voices to say ums and humans to not say um
when this is out, people are going to generate a clip of bigfoot and claim they saw bigfoot.
TH-cam shorts are filled with these now. They will get worse
Thats the best case scenario, ppl just having fun. The problem is when used for propaganda and crime. Not being able to trust anything a big societal problem.
Well maybe it backfires and then people will stop believing that stuff.
I can't wait to make fake UFO footage lol
I can't wait to make Shrek CCTV footage
The rock is called a geode
Thanks now I can sleep tonight
SORA KILLLER!!!
Until the next Sora, tomorrow.
"A rock that contains another rock inside" We usually call those geodes.
Funny thing about gen ai is that cloth smoke and fluids are so much more forgiving because they can deform in many valid ways and rigid objects are harder to model, while in cg it is the reverse as everything starts rigid until you simulate those dynamics.
The only reason why Google's AI are performing worse is because of censorship, otherwise they would've won by now. 😂😂
I don't know, Gemini Pro 1.5 is much more uncensored in AI Studio than GPT-4o or Claude 3 Opus, but it's still on the third place of chatbot arena.
@@nilaier1430 that's just keywords filtering, real censorship happens at the time of fine tuning the model.
Google has access to a lot of AI resources through DeepMind. The problem is fundamentally the execution.
WHOA 0:46 ! this is actually much better and on par with Sora ! can't believe it, how they improved upon their Veo that much in such a short amount of time? it's crazy good now, I'm so glad to see finally a true contender against Sora, respect to Google, lately they weren't much impressive with their AI endevours but this now is something else, I'm impressed
It's nice that the trees didn't move. It is odd though that some of them are growing in the middle of a driveway.
the AI GOD is almost here!
We call it Kevin.
just not to be disappointed again, otherwise everything is always good on stage and then
finally when you pray it will actually respond.
Can you go away please…🙄
1:49 - Whoa ! this looks so real ! now we're talking
Actually this was literally really really really cool. I guess.
Essentially yes. You’re really really cool and on point
One thing we DO know, is that this is basically really really really cool
@@WillyJuniorThat’s literally pretty pretty good and really really truly truly.
Notice: Don't have a drinking game for every time he says 'really'. You'll die.
That’s really truly truly good, literally.
Google has already lied multiple times about their demos
Dude needs a really really good thesaurus
RTX has been around in video games for years nothing new as you mentioned. It’s nice to see it’s being used in these models. The demos in the middle of the video were pretty standard in my opinion, but the rendering of the people with the sun were really good
From image to videos this good!!??..
Holy shit!.
>can generate
>generates stock footage of people moving their heads.
Could they at least set it on the moons of jupiter, or have the earth spinning once a second?
Lol, I hear you...but in their defense I think it's harder to generate mundane and "normal" looking scenes cos it's easier to see any mess ups
Man on horse realistic? Not a chance. Man turns to the left. Horse turns to the right. It looks like the man is riding side saddle on the away side of the horse.
{o.o}
Enjoy your ability to differentiate AI from real life, because it will be gone soon lol. Good luck spotting any errors when that times comes.
@@ToastyZach Oh, I expect it to come - sooner rather than later. I was a little surprised by the apparent visual lack of fit between horse and rider. To me it stuck out like recently hammered thumb.
{^_-}
i was looking for errors too like trees in the middle of driveways, and lone follow up waves not stopping at the rocks. But its getting way better very quickly.
Went back to watch the cowboy part turned up the brightness , i think its a matter of perspective, cos it looks normal when you can see his legs
They're not rocks, they're minerals.
Lighting in 3d adobe software like dimensions or even in photoshop is mastered . Still impressive how they put it into vid
The main thing you should point out is that its 8 seconds instead of 2 or 4.
And in regard to the time laps of the flower it might be nice to be able to change the lighting as it would over the course of the flower opening. For sometimes one may need to keep teh same lighting and other times not.
Geode. It’s a geode.
And continuing in my nit picking, for this is all truly amazing, (but one can always make things better as wel)l, is that in showing those reflection on the close up of Tokyo lights that they also model the Bragg angle that allows one to see the bottom of a puddle when the the line of sight is greater than the Bragg angle.
I need this Technology to research this.
It is a hypothetical dimension beyond the four dimensions(3 spatial dimension + time) that we experience in our normal reality.
It represents a state of expanded awareness and intuitive understanding, where one perceives the totality of existence and transcends the limitations of linear space-time.
It is not a physical reconstruction of reality, but rather a heightened state of perception and consciousness.
Ai I am not interested in shallow hypocrisy or deep fakes like these humans.
this technology is unrelated to RTX or raytracing or any rasterised or vector graphics in general. it's statistical matrix math, almost the exact same technology as LLMs but instead of predicting the next word it predicts the next pixel based on what it's seen before. it's guessing all the reflections and images completely, nothing is actually calculated, there are no rays that are being computed, it's all just fuzzy prediction not raytracing
with one caveat, they do train on videos of simulated raytraced environments to get more data. it really helps the model to have heaps of synthetic data in the form of pre-calculated raytraced videos done in renderers so that the model is taught what ray bouncing should ideally look like in a whole range of different weird situations, that makes it better at replicating the 'look and feel' of what reflections in puddles should look like, etc.
@AIGrid Can you try generate the Ferrets video?
openAI probably already has generative AI voice overlay onto in scene characters with generative ambient soundscapes and sound tracks augmenting their now 5 minute continuations of 3d video output with diffusion generated level of details zoomability to allow eye of god/ant real time perspective shifting in their sora 2 model lol
In regard to the video of the car going down the street one might be able to take Google Street view images and use this to fill in between those to make a smooth video as one "drives" down those real streets albeit virtually.
People like Emad want to give public access to such software.....FOR FREE internationally.
Good. Accelerate!
I am very skeptical on anything google announces until I see a live demo or use it in person, but what they showed looks pretty good
The issue is input control which could be through a 3D SOFTWARE app ie Maya,Blender etc...
plot change: they made a screenshot of a video
On one hand, the excessive use of slowmo is kinda cheating as you're basically having to generate less changes and just smoothly interpolate between very similar images; on other hand, it does serve the purpose of allowing the viewer to note a lot of details of quality of the generated image. Would be interesting if for every slowmo they also played the same clip in real time to see if the movements really do make sense; and play extended versions of the real-time that last the same amount of the slowmo ones in terms of watch time to prove they didn't just took the shortcut of stretching a very short generation.
They’ll release it after sora if open ai even drops that. Google is going to get embarrassed if they drop first.
It’s called a geod
This combined lot's of Google's researches.
I don't think it's a rabbit, I think it's a snake
They are afraid because there is no limits here, its incredible. The damage you can make is high.
Unfortunately until it's almost perfectly consistent, it's almost useless. Until then it'll be a cool toy, I'm excited to see the progress though. One step closer to simulated worlds we can all hang out in.
Looking forward to an open source model.
Googles is on too much shaky ground to care about their products.
2:50 The rays are actually coming from the right. That’s sunset light.
Really? That is not a rock inside another rock. That is a geode.
Questions are when and if released and cost and power required etc
“If this was something that was just coming out and it was state of the art, we would say this is very much so incredible”
Dear god! Nooooo! Stop! Think of the children!
this is cool but for actual film-making we still need consistent characters and speech
for now use reality?
@@davidvickers8425 reality limits us monetarily if we have a big vision
we bout to be cooked
Is it available? How much?
1:40. It’s an amethyst geode.
ray traced reflections have been a thing for years...video games definitely have puddle reflections
where's Paper Coral reef type of innovative prompts? what we see here is just everyday world, but Sora demoed truly innovative worlds come to life! Want to see more those type of comparison.
That rock things called a geode I believe
Everyone is making the same thing. AI market has become absolutely commoditised recently. I dont see the investments going well. Well. Atleast the magnificent 7 should float on well and good. Smaller startups tho... Lets hope for the best
I'm reluctant to believe Google demos right now. Gemini pro 1.5 is lagging behind. Nothing quite works the way promised.
absolutely terrifing
That’s solid
The rock is a geode.
'crystal ball' hahahahahahahahahha (>.
i think this video is really really made me think,
that i am very very aweare,
why this is really really gonna be,
really a very good example of
.....
a video with really really very very truly truly limited vocabulary
SORA killer... LMAO!! This is crap as is all what Google has recently announced.
Competition is good
Horse = Accordion
as cool as that is the videos might be more interesting and look a bit more realistic if the facial expressions change.
Thus is a person is smiling or laughing at the start of the sequence maybe they smile fads over the course of teh scene or the laugh turns into just a smile.
This would give more nuisance and make it look a little less computer generated.
Disappointed in this video because the information is old, we can't actually use it yet and there is no information about when we will
this is crap compared to SORA, is not very very very good nor really really impresive
Im shocked. Either way competition is good
Really, really
Until AI stops making creepy blob humans consistently, I'm not interested
Did you watch the video?
@@davidvickers8425 some of it
the horse seems amazing but it is not. everyone is falling for it. there is a perspective problem when the horse moves the head. the outline trick is in fact tricking all "not so smart eyes". no offense of course. but still if used in the right contest is pretty cool!
I would hope its not perfect yet. Its gonna shake the planet
have you played red dead redemption 2 at all? Or any of the big name hunting games. This shouldn't be too difficult if you judge by that.
Way too much stuff, being announced but can't use nothing. This is so horrible.
shit crazy
This is clearly not as clean as sora
This is not good. Easily used for bad intentions
homie doesnt know what a geode is..
Scraping the internet, copying the art, photography, film and writing of people who have worked on their skills for decades.
It’s hard to do because it doesn’t have any bones? Okay dude, whatever you say.
say less man. In 3d animation they attach a skin to deform according to how it is attached to the rigging(bones) a jellyfish is hard to animated because you have to rig it differently to a biped or quadraped, insect bird etc. It doesn't look realistic because it would look like an umbrella opening. He's impressed by the natural look of the jellyfish animation.
@@davidvickers8425 AI doesn’t discriminate between these things, only people that don’t understand think otherwise, like the guy making ai videos
You’ve hit rock bottom dude. This is it. You can’t go any lower than this
tf you on about
@@edwinungerer7989 it’s called having an education
@@ollybreh95 tf you on about?
@@edwinungerer7989 bless you
@@edwinungerer7989its a geode pun i think
Looks lame.
Meh
how?
You say lots of words, but convey so little information
wut
harsh review i liked this video
Google’s, not Googles. Yes I’m only here to dunk on the video. I know before even watching that it’s going to be total crap, read by someone who genuinely has a stroke every time they try and string a sentence together.
You talk total rubbish
wut?
You love saying actually 🤮