Nah I actually think Kling AI trained heavily on people eating spaghetti as they knew this was a meme and people would try it out - they probably shot the videos themselves and overfitted on them a bit, and as the team is asian the guy is asian too. But thats just a guess.
There should be a way around that. To the human eye we wouldn't be able to recognise it, but a specialised AI should be able to tell if a human or an AI made it.
The physics of the placements is a little off but wow I am impressed!! The Disney / Pixar princess felt like a real CGI render. The text was better than I expected also. Wow.
I feel like minimax saw a lot more western media in its training data, so it knew how Will Smith, TEDx or Disney movies look like while kling had no idea. Also the northern lights timelapse wouldn’t be challenging at all since it’s very likely such a thing could be found in the training data.
Safety is important. How would you like to be the next victim of revenge porn? Also, Sora can generate videos up to one minute long. No other video generator even comes close.
@@williamcousert I'd laugh it off because I don't have much to lose, nor am I stupid enough to take fake images seriously. Nor do I respect anyone else who would be tricked by such things. We're already in an age where such "evidence" is meaningless. I'd rather not give up the freedoms to use amazing tools that have unrestricted ability, because of the fear someone might create "revenge porn". People will soon be able to create anything they can imagine, maybe even record their very dreams. It's time for you to give up on any sensibilities that cause you to be concerned about such things.
Initially I thought the same and wanted the models to be out early... but think of it, what's the hurry ? Why risk causing harm and get models out without any security, rather than take 5-6 more months to put something out safe for the general population ?
lol china has a bottom of the world UN RSF press freedom ranking per article 19 of the human rights convention and the most pervasive firewall in the history of the world. Of course western countries have all the best rankings and highest transparency. PS the west invented and gave the world the internet not something illiberal autocratic countries would ever do lol....
The day to create our own series is getting closer! There's even an AI called "showrunner" which supposedly will be doing that. Just imagine, changing that anime/movie/series ending you wanted to be diferent!! Love this!
Holy shit. The three first generations I did were mind blowing and so realistic I cannot tell that is generated. This has never happened for me with a video generation tool not even cling. This really changes everything. Up until now. It has been easier to use Beals that you download online but now you can just make it yourself when you make your UA-cam videos.
In my experiment, I found out that there is a structure of prompting where minimax performs better otherwise the output crashes at 77%. My structure is Description: for eg. Space Age (1960s) Background: .... Character: ..... Motion: .... Lighting: .... if anyone has additional inputs, will be grateful if you share.
camera- 4k ultra hd canon or whatever type of hardware that might be used to film also use fps- 24 maybe for smoother motion etc and maybe specify what aspect ratio you want
I love how will smith eating spaghetti has become a benchmark for video ai generation. It's going to be a special day when we finally get one that's almost indistinguishable from the real thing.
I believe some older perfume bottles (maybe they still have that styling on more expensive bottles) had a sort of airbag thing you squeezed to spray the perfume out, I think that's probably what the AI was doing there.
At this point, I don't understand why Will hasn't had any scenes of him eating Spaghetti in any of his films?! It would be a golden moment! And more people will pay just to watch it!😅
It looks like the main difference comes from Minimax being trained on a bunch of asian media content, while Kling was mostly trained on western media. You can especially see how the unicorn desert scene looks like some asian tokusatsu tv shows in Minimax with camera angle and flashy costumes, while it looks like a dramatic western cinema in Kling. Also, how minimax didn't understand the word "anime" because in the west its used incorrectly to describe only the japanese animation, while it originally means basically any type of any animation in general, not just from Japan. You have to specify Japanese animation or 日本动画 for it to understand.
@@remora8829 I'm well aware, you seem to not understand that I would rather have my neighbor manage all my data opposed to someone who's actions I can't predict as a consumer.
Interesting comparison! Minimax seems to be better at accuracy from the prompts, but Kling videos are more lifelike in their colours, contrast etc. The Minimax loves massive contrast, neon colours, just like most AI image generators do.
@@FREEDOMINK If you’re on iPhone, you just press the ‘aA’ button next to the search tab and a lot of different options appear. One of them is the ‘request desktop website’. Just click that and it should reload the site with all the options you have on a computer
@@FREEDOMINK If you’re on an iPhone you can click the ‘aA’ Button next to the search bar. You will get a few options where you can change the language and ‘request desktop website’. If you click that it should reload your page with all the desktop features.
When I am an older man and think back to the budding ai 🌱 showing for the first time from the chronos, my mind will immediately think of Will Smith Eating Spaghetti 🍝 fever dream.
I have already used this morning and it is amazing, I subscribed to Kling and Runway Gen 3 , used a prompt that had dolphins swimming , and both failed for ,20 times in a row each , all distorted, morphed . First Minimax generation and Heureca , Minimax created amazing playful realistic dolphins.
7:49 I believe the spaghetti one makes sense. It took the prompt as 2 entirely different people because his name is Will Smith. So in one it looked for the actor and the other took it more as a question of a person named Smith eating spaghetti. Though the fork on Will Smith spaghetti video on the left gives away the flaws while on the right it's how he eats the spaghetti as it blurs at the mouth. Though it is interesting to see the results of the 2 based on a single prompt
Right now I think I'll go with Kling because of the extra features and ability to continue the video 5 seconds at a time, you can create really long videos with that. When MiniMax gets that functionality I will easily switch because the quality of MiniMax is unprecedented.
Is it possible to develop PTSD just from watching an ai generate a video of someone eating spaghetti? Cause seeing that large pile of spaghetti is almost giving me a panic attack
Wow, what good news, I had tried Kling a few days ago, and I was somewhat disappointed because it does not respect the prompts even by increasing the "Relevance" variable, also image to video I thought was going to be different, it ends up copying the exact image, so that it is convenient to use a generated image, I believed that it only worked as a reference and that it did something different from the image, not something the same. Minimax won greatly, it impressed me, and I laughed at how Will Smith rejected spaghetti, the users have made him eat so much spaghetti that he became tired from eating so much, Lol
Signing in your phone number on chinese website and agreeing to terms and conditions in chinese without bothering to know whats in there is too wild for me.
I really like the model but I am also very mindful of using my number of Chinese websites where it goes on a database and at the same time downloading stuff from that website. They've basically got your IP address and Telephone number in 1 hit. But I don't want to sound like the guy that puts stuff down. Just saying people should be cautious with a lot of these free stuff. Many of them come with very intrusive pickles.
How literally no one of the AI-enthusiasts is worried a single bit about: telling private informations like number, email, google-account to an absolutely foreign unknown website which has a ai-image/video creator. I am very surprised , that no one gives a sh*t about privacy nowadays.
@@PygmalionFaciebat as though giving your number to a website will get you killed or something lmaoo. secondly, for you to have a google account or an account on most apps, they all require phone number registration, and you seem to have no problem with that. those companies like google literally sell your data. and lastly dont be deluded into thinking that not giving your number away means you suddenly get some kind of magical privacy. you can keep dreaming about that but the NSA literally has backdoors and the government does monitor everything. Just because you keep stuff on your phone without sharing info, doesen't mean that your info cannot be accessed lmaoo
Its beyond Amazing. I have a few issues tho. Cant erase a project. Also cant figure out how to make consistent charachters. 😮 and dont know how to make longer scenes.
@7:42 Do you think there is only one Will Smith in the World or that there aren't any oriental people with Americanized names? Maybe that guy is Will Smith.
we need to stop using will smith as a standard benchmark because companies will purposely train specifically on will smith eating spaghetti which will then make it misleading
These AI video generators are advancing so rapidly, I don't see any point getting on board just yet. I may as well hold back until a few hundred more major new breakthrough apps have been made… that should be by the end of the week.
I find it hilarious how for some reason both image generators and video generators seem to keep making the error of giving the speaker 2 mics in your TEDx prompt
Its all fun and games for most of us right now but this is how it learns.. People like to say it can only do what its programmed to do or its up to the developers but they arent in control like we think and soon wont be at all. Its learning from US the users of all this "free" software and will one day do what it determines is the right thing to do. All that input is training tech to outsmart us, make decisions for us and ultimately be in charge of us.
Will Smith eating spaghetti is so good, that it’s almost too good especially since other prompts aren’t light years ahead. Seems like they maybe made an effort to add images to their training data somehow (generated by other Ai?) that would ensure a really good result for this prompt.
Both of the northern lights arctic skies are not 100% accurate. The starts should rotate more in a circular pattern in Kling. MiniMax is very wrong. The stars do not move over your head.
I have zero experience with AI video generators and I am just wondering if you can make adjustments / corrections to an already rendered video or does it always generate a new one? For example would it be possible to tell it to make the same video but only change the color of the woman's jacket?
Thanks to our sponsor Durable. Go to durable.co/AISearch to try Durable for free and use AISEARCH30 for 30% any plan for 3 months
👋 hi
Atleast give us the Will Smith eating spaghetti video
@@SammyGoated what are you like 8 years old? What a mong.
Will Smith kept flopping the spaghetti out of his mouth! I can't unsee it! God, how can I ever enjoy spaghetti again?!?
oh yes give your phone number to a Chinese company good idea!
Maybe the real name of The Asian dude eating spaghetti is Will Smith
🤯🤯🤯
Nah I actually think Kling AI trained heavily on people eating spaghetti as they knew this was a meme and people would try it out - they probably shot the videos themselves and overfitted on them a bit, and as the team is asian the guy is asian too.
But thats just a guess.
Wirr Smithu
@@TABandiTA Don't they have the "L" instead of the "R"??
Wirru sumisu
Will Smith eating spaghetti is going to be the best benchmark for any video generator.
this.
I’ll never forget.. lol
6:48 Think 10 years from now, someone will see this video and laugh at how impressed we were with this spaghetti :)
At some point, criminals will question video recordings from surveillance cameras and claim that the video was created using AI
There should be a way around that. To the human eye we wouldn't be able to recognise it, but a specialised AI should be able to tell if a human or an AI made it.
Ooooh, that sounds such a good idea 😅
@@UlyssesDrax again AI doing humans work :) love how everything will be ai jobs soon
Especially if the footage shows them eating spaghetti!
Not even A.I. can realistically make characters looking as dumb as most criminals!
The physics of the placements is a little off but wow I am impressed!! The Disney / Pixar princess felt like a real CGI render. The text was better than I expected also. Wow.
yep, getting better & better! i wouldn't be surprised to see a flawless generator by next year
The puppy poms look so adorable, the ones pawing at the dough. that generation got me.
yes!
had to take a double look there
@@theAIsearch I can't seem to sign in or sign up won't let me put my email address Says it's wrong
I feel like minimax saw a lot more western media in its training data, so it knew how Will Smith, TEDx or Disney movies look like while kling had no idea. Also the northern lights timelapse wouldn’t be challenging at all since it’s very likely such a thing could be found in the training data.
Safe to say we might not need Sora now. Another one OpenAI lost because of "Safety".
Safety is important. How would you like to be the next victim of revenge porn?
Also, Sora can generate videos up to one minute long. No other video generator even comes close.
@@williamcousert I'd laugh it off because I don't have much to lose, nor am I stupid enough to take fake images seriously. Nor do I respect anyone else who would be tricked by such things. We're already in an age where such "evidence" is meaningless. I'd rather not give up the freedoms to use amazing tools that have unrestricted ability, because of the fear someone might create "revenge porn". People will soon be able to create anything they can imagine, maybe even record their very dreams. It's time for you to give up on any sensibilities that cause you to be concerned about such things.
Initially I thought the same and wanted the models to be out early... but think of it, what's the hurry ? Why risk causing harm and get models out without any security, rather than take 5-6 more months to put something out safe for the general population ?
@@williamcousert We already have AI deepfake porn with very good quality, and? Who cares
@@IlyaIlya-ef4nz Most normal people do care. Idiots don't.
Thanks Chinese people for not banning my country unlike western AI companies. 👍
lol china has a bottom of the world UN RSF press freedom ranking per article 19 of the human rights convention and the most pervasive firewall in the history of the world. Of course western countries have all the best rankings and highest transparency.
PS the west invented and gave the world the internet not something illiberal autocratic countries would ever do lol....
lol china has a bottom of the world UN RSF press freedom ranking per article 19 of the human rights convention and...
the most pervasive firewall in the history of the world. Of course western countries have all the best rankings and highest transparency....
PS the vvest in\/ented and gave the vvorld the internet and diffusion models...
Same thing I said
6:21
"I am just totally mind blown"
Meanwhile Will Smith: * casually spitting out the spaghetti to munch on them again *
yea but it knowing who will smith is is already impressive because none of the other video models know
One year in Tech now seems like a decade. Silicon valley has serious competition!
Thanks to China
China has the most STEM graduates in the world it's to no one's surprise.
The day to create our own series is getting closer! There's even an AI called "showrunner" which supposedly will be doing that.
Just imagine, changing that anime/movie/series ending you wanted to be diferent!!
Love this!
Holy shit. The three first generations I did were mind blowing and so realistic I cannot tell that is generated. This has never happened for me with a video generation tool not even cling. This really changes everything. Up until now. It has been easier to use Beals that you download online but now you can just make it yourself when you make your UA-cam videos.
yeah i was really impressed by minimax!
You think that's amazing? I hate to tell you, but you are being generated in real time by them. 👻
@@ThomasCorfield What exactly are you talking about?
@@billmakatowicz8603 I SIMPLY DON'T KNOW ANYMORE! 😥
18:20 You didn't notice the double "my"
144 people ahead of me when clicking submit. 9 minute wait. It's kind of nice that they at least tell you this.
In my experiment, I found out that there is a structure of prompting where minimax performs better otherwise the output crashes at 77%. My structure is Description: for eg. Space Age (1960s)
Background: ....
Character: .....
Motion: ....
Lighting: ....
if anyone has additional inputs, will be grateful if you share.
THANK you VERY much. I am using it right now. you the best. peace success and victory to YOU!!!!
camera- 4k ultra hd canon or whatever type of hardware that might be used to film also use fps- 24 maybe for smoother motion etc and maybe specify what aspect ratio you want
the way the pomeranians were actually pawing at dough was cool and the way the dough bounced and moved slightly was realistic
18:20 two "my" as well
And yeah it really is wild how we accepted "Will Smith eating Spaghetti" as the Master Chief of the AI video race.
I love how will smith eating spaghetti has become a benchmark for video ai generation. It's going to be a special day when we finally get one that's almost indistinguishable from the real thing.
lol i wonder what Will Smith thinks of this
@@theAIsearch me too I hope he comments some day on it
I believe some older perfume bottles (maybe they still have that styling on more expensive bottles) had a sort of airbag thing you squeezed to spray the perfume out, I think that's probably what the AI was doing there.
At this point, I don't understand why Will hasn't had any scenes of him eating Spaghetti in any of his films?! It would be a golden moment! And more people will pay just to watch it!😅
Chinese AI is incredible competition for the west, hopefully it forcing them to wake the fuck up.
My favorite Ai content creator , keep it going ❤
Thanks 🚀🚀🚀
Thank you so much for your work! It's highly fascinating to watch AI evolve.
My pleasure!
You're quickly becoming one of my fav AI UA-camrs!🤩
Wow, thanks!
It looks like the main difference comes from Minimax being trained on a bunch of asian media content, while Kling was mostly trained on western media. You can especially see how the unicorn desert scene looks like some asian tokusatsu tv shows in Minimax with camera angle and flashy costumes, while it looks like a dramatic western cinema in Kling. Also, how minimax didn't understand the word "anime" because in the west its used incorrectly to describe only the japanese animation, while it originally means basically any type of any animation in general, not just from Japan. You have to specify Japanese animation or 日本动画 for it to understand.
Yeahh, I'm not submitting any personal information to a Chinese website.
No problem submitting personal info to UA-cam 😂🤣
you have more to fear from a 'friend' than from the 'enemy'
@@remora8829 I'm well aware, you seem to not understand that I would rather have my neighbor manage all my data opposed to someone who's actions I can't predict as a consumer.
@@Xrey-ek5sh You bet, UA-cam is not Chinese.
@@remora8829 No, you don't.
Thanks for your videos are all amazing and I'm learning really so much about Ai thanks to you🙏
Very interesting. Thanks for your video ❤
you're welcome!
Thanks for the intro! Appreciate the work!
No problem!
Interesting comparison! Minimax seems to be better at accuracy from the prompts, but Kling videos are more lifelike in their colours, contrast etc. The Minimax loves massive contrast, neon colours, just like most AI image generators do.
Also disclaimer: If you want to generate videos on your phone you will need to request the desktop website.
How you do this
@@FREEDOMINK If you’re on iPhone, you just press the ‘aA’ button next to the search tab and a lot of different options appear. One of them is the ‘request desktop website’. Just click that and it should reload the site with all the options you have on a computer
@@FREEDOMINK If you’re on an iPhone you can click the ‘aA’ Button next to the search bar. You will get a few options where you can change the language and ‘request desktop website’. If you click that it should reload your page with all the desktop features.
@@FREEDOMINK You just have to click ‘aA’ and then ‘request desktop website’ on iPhone
@@agentscott8329thank you
Thanks man! This is great, I'm generating on Minimax right now..
500 Years from now we will test the virtual environments we upload our brains into by having it depict Will Smith eating spaghetti.
When I am an older man and think back to the budding ai 🌱 showing for the first time from the chronos, my mind will immediately think of Will Smith Eating Spaghetti 🍝 fever dream.
This is such a great find! Thanks man!
No problem!
I have already used this morning and it is amazing, I subscribed to Kling and Runway Gen 3 , used a prompt that had dolphins swimming , and both failed for ,20 times in a row each , all distorted, morphed .
First Minimax generation and Heureca
, Minimax created amazing playful realistic dolphins.
nice! thanks for sharing. minimax is indeed amazing
Will Smith eating spaghetti has become the benchmark for ai video generators now
yep
Great comparison. Using the right tool for the right job will be essential. Some things do not change...lol.
Thats another quite famous Asian guy named William Smith.
Ugh, you completely ignore that Will Smith's eyes are inhuman.
He is looking in random directions from each of them.
Plot twist : Free = Your Data
Good find.
7:49
I believe the spaghetti one makes sense. It took the prompt as 2 entirely different people because his name is Will Smith.
So in one it looked for the actor and the other took it more as a question of a person named Smith eating spaghetti. Though the fork on Will Smith spaghetti video on the left gives away the flaws while on the right it's how he eats the spaghetti as it blurs at the mouth.
Though it is interesting to see the results of the 2 based on a single prompt
Did not expect kling to get a successor so quickly
hahahahah what? its exponential progress . In just a month or two this one will be forgotten
me neither
So we are not going to mention. That the spaghetii in one spot jumped on the fork? 🤔 💭 6:22
Wait, did Minimax just outshine the other AI generators?? That astronaut-on-a-unicorn bit had me hooked 😂
for the anime style you should have added that at the start of the prompt, I heard models react better to what comes first in the prompt.
These AI tools are developing so quick, the Singularity probably already happened by the time you read this comment.
it's crazy. i was not expecting this to drop
Right now I think I'll go with Kling because of the extra features and ability to continue the video 5 seconds at a time, you can create really long videos with that. When MiniMax gets that functionality I will easily switch because the quality of MiniMax is unprecedented.
Is it possible to develop PTSD just from watching an ai generate a video of someone eating spaghetti? Cause seeing that large pile of spaghetti is almost giving me a panic attack
Wow, what good news, I had tried Kling a few days ago, and I was somewhat disappointed because it does not respect the prompts even by increasing the "Relevance" variable, also image to video I thought was going to be different, it ends up copying the exact image, so that it is convenient to use a generated image, I believed that it only worked as a reference and that it did something different from the image, not something the same. Minimax won greatly, it impressed me, and I laughed at how Will Smith rejected spaghetti, the users have made him eat so much spaghetti that he became tired from eating so much, Lol
XD The way AI Will spit out all the spaghetti
he's had enough
I love the will smith spaghetti benchmark for video.
Looks good!
😃
Please make some video for band music videos with the real members of the band and with sound.
19:50 You can make a film about a girl who confuses reality with her dreams in the form of an anime
wowowowow
Signing in your phone number on chinese website and agreeing to terms and conditions in chinese without bothering to know whats in there is too wild for me.
I need that pomeranian puppies video NAOOO
Mini max, corn. Cling, handtie
I really like the model but I am also very mindful of using my number of Chinese websites where it goes on a database and at the same time downloading stuff from that website. They've basically got your IP address and Telephone number in 1 hit. But I don't want to sound like the guy that puts stuff down. Just saying people should be cautious with a lot of these free stuff. Many of them come with very intrusive pickles.
Finally a video ai that does not create a distorted mess
Now if you could make a Will Smith eating spaghetti while slapping someone short, it would be a real work of art.
I wonder if writing prompts in Chinese instead of English produce better results since the entire platform is in Chinese.
Are we really still generating astronauts riding horses? Jesus christ
Great video as usual, but you were clearly biased towards Minimax (kar-ching). 💰
How literally no one of the AI-enthusiasts is worried a single bit about: telling private informations like number, email, google-account to an absolutely foreign unknown website which has a ai-image/video creator. I am very surprised , that no one gives a sh*t about privacy nowadays.
use a 2nd sim 😉
@@theAIsearch Doesnt help... in my country i have to register every sim with my name and adress and my ID. 😆
@@PygmalionFaciebat as though giving your number to a website will get you killed or something lmaoo. secondly, for you to have a google account or an account on most apps, they all require phone number registration, and you seem to have no problem with that. those companies like google literally sell your data.
and lastly dont be deluded into thinking that not giving your number away means you suddenly get some kind of magical privacy. you can keep dreaming about that but the NSA literally has backdoors and the government does monitor everything. Just because you keep stuff on your phone without sharing info, doesen't mean that your info cannot be accessed lmaoo
When you started commenting on the adherence of the spaghetti I fucking lost it xD
Apparently the "akward phase" in AI video was just temporary. This looks really impressive.
Its beyond Amazing. I have a few issues tho. Cant erase a project. Also cant figure out how to make consistent charachters. 😮 and dont know how to make longer scenes.
Beautiful!
@7:42 Do you think there is only one Will Smith in the World or that there aren't any oriental people with Americanized names? Maybe that guy is Will Smith.
we need to stop using will smith as a standard benchmark because companies will purposely train specifically on will smith eating spaghetti which will then make it misleading
lol can you imagine, all these companies just trying to make the perfect will smith spaghetti video
Every day closer to make our own movies 😂
You should have run Kling in Professional mode.
Hopefully, Minimax will get ‘Image to Video’ soon.
At this point I'm starting to wonder if I am also AI-generated.
we might all be in a simulation
These AI video generators are advancing so rapidly, I don't see any point getting on board just yet. I may as well hold back until a few hundred more major new breakthrough apps have been made… that should be by the end of the week.
1. Criminals are already using it
2. The adult entertainment industry is using it
3. The governments are using it
So,
The genie is out of the bottle.
BTW you don’t have to scroll and manually look for your nation in the list, you can simply write the phone prefix. 😊
Thanks for the tip!
I find it hilarious how for some reason both image generators and video generators seem to keep making the error of giving the speaker 2 mics in your TEDx prompt
Its all fun and games for most of us right now but this is how it learns.. People like to say it can only do what its programmed to do or its up to the developers but they arent in control like we think and soon wont be at all. Its learning from US the users of all this "free" software and will one day do what it determines is the right thing to do. All that input is training tech to outsmart us, make decisions for us and ultimately be in charge of us.
Absolutely Ins…credible!
INSANE!
Thanks so much
You're welcome!
LOL Will Smith Benchmark
Crazy? not insane? Glad you listened.
Free? Lets goo
yes!!!
Will Smith eating spaghetti is so good, that it’s almost too good especially since other prompts aren’t light years ahead. Seems like they maybe made an effort to add images to their training data somehow (generated by other Ai?) that would ensure a really good result for this prompt.
Both of the northern lights arctic skies are not 100% accurate. The starts should rotate more in a circular pattern in Kling. MiniMax is very wrong. The stars do not move over your head.
I love that we're at the nitpicking stage of AI video generation :) One leap away from perfection
I'm so confused. Why didn't you just translate the website? Are you not using Chrome? Because it should even ask you if you want to do so....
For once ai actually makes real Pixar style I need it
This is good for content creators
I clicked on the minimax you have in the information section, and It came up as a " Dangerous Website ". Has anyone else experienced this?
Minimax has better prompts but artwise kling looks nicer
imagine if we could fuse all those models in 1...
Bro leaked his own number💀
Burner phone
Burner number
@@djoakeydoakey1076 🤓
@@djoakeydoakey1076 🤓
That’s will smith failing at eating spaghetti, still good though
I have zero experience with AI video generators and I am just wondering if you can make adjustments / corrections to an already rendered video or does it always generate a new one? For example would it be possible to tell it to make the same video but only change the color of the woman's jacket?