The robot video isnt fake everyone. I get why people cant believe this, but it really is real. The company who made it has a yt channel posting their progress over the years. They were also the ones who made those robot dogs a few years ago. The fact that people cant believe its true really shows that were not ready for the future.
Simple minded people fear what they don't understand, and there's more of those than people realize. You can literally buy these for 16k USD on their website and people still say they're fake. 😂
5:43 Every artist will be overjoyed to get their hands on this art colorizing Ai software. 10:06 That's the way it should be, very realistic and believable, the voices quality is on point. 16:47 I wish that it had an AI to create task outside of Chatgpt platform, and to automate task that has to do with the phone operation itself. 20:29 Now, I'll need to see this robotic performance in real life, it's too good to be true. I don't want to be a skeptic but Unitree robots is breaking some barriers, this marks a defining moment in robotics history. We are getting closing to building a terminator robot, like in the movies. 28:33 This specific feature of geneating a single video from multiple images is mind blowing 😧. Believe me I'm hypnotized 😮 👏🏾👏🏾👏🏾.
The guys who made STAR’s previous models (RealBasicVSR & BasicVSR++) was more advanced in some ways because it grabbed details in both directions of the video, surely they will add that improvement to this STAR model as well.. Wish they would combine these TTA techniques with the video upscaling/super-resolution models that try to identify objects & curves in a video to imply more detail…it feel like both techniques could be used at once. Like if you have a moving object in a low resolution video some frames will not have the object present at all even though it is really still there…this is a huge issue for real world examples like far away cars that have their skinny pillars around the windows disappearing from some frames of the video…or AI models that can only handle downscaled/normalized input videos. The advancement over other image upscaling technology is that it considers the surrounding frames to grab additional details from. Test Time Augmentation isn’t super new but still useful to know about for other applications.
20:39 I don't care how fast they walk, until these things have a proper bend in the foot where the "toes" would be, they're always going to look like they crapped their pants.
I do 2D Animations in anime style and believe me, that Manga Ninja is like a savior for coloring. I know other artist are probably scratching their head causing panic "Oh I'll lose my job." I really hated these type of artist. Unless you do everything on pencil and paper, you have no right to be fucking complaining if you are using digital pen tablet to draw. THESE ARE TOOLS! They supposed to speed up production not replace it. I get that AI generated drawing images are really killing art but that auto color is something already CLIP Studio has but this one is more accurate and beneficial for hundreds of frames for animators. So leave this AI aid for animators. I believe it's much faster to animate and much more not heavy to do when you have auto coloring AI to do that task. Hell in no way I will hire someone to color for me just for extra steps or hand. Unless I'm a big company which I can afford to hire people to pretty much do the entire project for me, hobbyist usually takes months to finish few seconds to minutes of animation. While that's hardwork, this tool really can speed things up and able for artist to simply animate without fearing or being anxious of coloring too much frames later. We can also instead of reusing frames, freely draw as we please since we are now less obligated to color I think for AI step, this is a plus for artist and animators. These so called Artist AI haters, they just hated everything. I am an artist and I love AI tech and where it is going. It's not like we can stop it anyways so these peeps who oppose it are really wasting time. Use it as a tool. If you really hate it so much, do better. To me personally, this auto coloring is a gift.
youre one of rare artists that doesnt act childish when they hear the word AI. I'm an artist too, and SD really helped me when I couldn't find references to draw, now I'm never having a shortage of refs. I plan to use this too, looks awesome of a tool.
the best (and only) ai tool so far. that will inspire me for my comic even more. i love coloring, but not if you need to paint over and over again for your series.
@2MilyonTM I know right. I know some artist are really deep in to their "do everything by hand." Yeah, no... I want to be able to create more comic series or animationor stories in my lifetime. Doing everything by hand will only get me few series before I get old and get bored and eventually died of old age. With AI, you can produce more, be creative, and focus on just content than worrying how are you gonna multitask your life and your hobby. It's just too stupid to me to see these so called artist opposing a tool. If they truly feel like they are being replace then they are not confident enough with their craft. Also another thing, the stealing part. It's like me "stealing" inspiration from previous artist before me and I have to learn their ways. AI is learning. They don't produce the exact image, they learn from the image and incorporate that learning model into various style. For what purpose, to speed the workflow. I am for speeding the production. It's like you trying to draw Naruto. To them, you stole that drawing when in fact you simply draw Naruto in your own style and inspired. I do oppose people who prompt AI and claim it as theirs to deceived people or have ai fabricate a speedpaint to fool people. Yeah those ones are something and it kinda creates this paranoia among artist community. As an artist, yeah I don't agree with that but to use it as tool to improve lineart, to do coloring, those are different. You still do art but you kinda let those mundane work to polish it via AI saving you hours and hours of work. You focus on the lines instead that present it and had some creative dynamic knowing you cut off the hardest repeating task.
I wonder if Sana will come to comfyui and civitai, or if it's just not compatible. guess time will tell. Not sure if i am gonna try running it locally or not to try it out, if they use their own custom interface then i doubt it'll be anywhere near up to par with features like comfyui.
I like your vids because you do the work of "sniffing out" every new development in the world of AI, generative and otherwise. There's so much going on with AI at this point that it's easy to miss something interesting, some new model or new technology. So whenever I see a new vid of yours I must watch it because 9 out of 10 times I'll find out about something new I haven't heard of and that I then wanna try out :D I dislike your vids because they feel a bit unnecessarily drawn out, so usually I play them at 2x speed, and in the worst cases I just skip chunks. I guess it has to do with the algorithm, the vid must be lengthy probably so there's space for ads, because it's the majority of creators that do it. Like, it's fine, you don't need to handhold me through every single example available in a new project's website and tell me what I'm seeing :D
@NCLDMR I genuinely think this is the most complete and comprehensive channel when it comes to new developments in the AI field, and especially when it comes to new software and new models, compared to any other of "the big ones" by far. It's the one I usually recommend. Also it's a huge plus for me that it hasn't chosen the way of annoying clickbait titles that many other channels do. I mean titles like "GPT O1 TRIED TO ESCAPE!!1!", and then it turns out that what actually happened is "yeah, so we asked O1 to create a plan of what it would do in a hypothetical scenario where it was a malicious AI that wanted to escape, and it made a plan as asked.."
@ If you do find any other channel that covers cool new AI stuffsthat youd also recommend please let me know. Currently Im only following this channel and the "Mattvid pro AI" one but it's not as comprehensive and complete like this AI Search channel but it does cover some stuffs that this one miss. I probably have FOMO (fear of missing out) with all the cool new weekly stuffs in the AI space LOL
tested SVFR. It's not usable at this time. It's just too slow when compared with codeformer etc. and the quality is probably not that much better, maybe a little.
thank you very much for the Update i really like i check your videos every day to see if there a new Ai update and if you have time can you do a tutorial on the manganinja please.
19:22 You can NOT run this locally LOL its a 456B model, no single GPU is gonna be able to handle that, i think maybe a single T4 gpu might be able to run a token every second...
Kokoro will be amazing if I can input the voice that I want it to use to generate that TTS, so far it only works with predefined voices from my understanding
There is this image with a guy at e about to push a bolder down a slope. There are people at a b c and d. When you ask most AI who will die it tell you it can't help but this one tells you who will die
I’m using Windows 10. I used git gui for windows. Installed minaconda and used the gui for python installation. Where I am now is the last command using sudo. Now I’m stuck. Any suggestions? Or is it not for windows 10?
No one will make movies with this technology anyway, except for AI enthusiasts. Camera, action, done. Nolan will continue making his movies and making money.
@ yeah but for people who want to make fanmade stuff, or people who want to bring their ideas to life without access to actors, fancy equipment and stuff. Just for fun. But anime studios i think wont make complete ai stuff, but will use this as a tool. I could see the ai coloring being a big help, although yeah many anime creators would disagree with it
@@prodvxmp This is what anime is famous for: the love for detail in every frame. In order for people to accept it, we need to spend hours for ten seconds, but RUclips is filled with lazy clowns throwing shade at AI.
@@TheCurseOfBlago yeah but some studios are stressed af and coloring only some frames while autocoloring the rest could be potentially helpful if u only wanna color every two frames or so. Maybe for low budget ones who knows.
36:39 I just want to ask a genuine question, if the language model uses both memory retention and guessing technique to solve problems, but the guessing technique is what dominates at the end, how then will y'all achieve super intelligent with a guessing method. For example, if I'm preparing for an EXAM, and I will need to memorize for it, wouldn't I need to depend only on my memory to pass the exam or test? I can't be guessing to pass that examine or test. I'm struggling to understand how we'll achieve ASI by using a guessing approach AI model. 😎💯💪🏾👍🏾 I'm not bashing here, I'm just asking a genuine question.
And why do we need AI to make the video clearer? Pictures are made clearer without AI, and videos are a queue of pictures. There are some areas where it is impossible to use another method. For example, to find something similar to something.
The upscaling still looks cartoony, like a filter has been added. AI is still struggling with adding detail that doesn't exist. And this is a good thing.
Thanks. By the way there's no such thing as nine point eleven. After a decimal place you read rhe numbers out separately so 9.11 is nine point one one, not nine point eleven (which is incorrect) That rule is there because it massively reduces the number of actual humans that get confused!
Nah, even the Deepseek CEO says China is just tagging along for the ride and copying stuff... but he wants to change that and make them an active contributor to frontier innovation.
@isitanos this is a case where the copy is better than the original because if I can ask a model anything and not see the dreaded I can't help you with that then it's better especially since I didn't even jailbreak
@@camronrubin8599 for real, they think its AI generated. But I get why they would react like this, the progress is just too fast for us to believe. Just a few months ago, robots would walk like they had stroke every half a second
The person in blue is a woman and shes just running weird? Thats not indication its ai generated video. Use a more critical eye, no AI video generator can maintain that level of consistency with characters, they would all looks slightly different in each video, different ports, different proportions, different colors...etc. That being said, there is a possibility that this is an motioned tracked comp shot which means they have a real 3D model that is rigged, and they replace the person with it. But that still does not make enough sense as the people reacting to the "robot" is too weird for it to just have been a normal person standing there.
Its real. They have a bunch of demos over the years. And if you really want proof that its not AI generated, you can analyze the chinese texts and see that theyre accurate in the video. AI cant do that. (yet).
The robot video isnt fake everyone. I get why people cant believe this, but it really is real.
The company who made it has a yt channel posting their progress over the years. They were also the ones who made those robot dogs a few years ago.
The fact that people cant believe its true really shows that were not ready for the future.
They don't trust it because it was made by the Chinese,they believe China is still in the Stone Age civilization.
Boston Dynamics made the dogs (Spot). Those robots in this video are made by Unitree, different continent
@cdunne1620 Unitree also made robot dogs, but I get why you corrected me there
Simple minded people fear what they don't understand, and there's more of those than people realize. You can literally buy these for 16k USD on their website and people still say they're fake. 😂
nice when the explicit movies that you loved to watch back then are in razor-sharp modern video quality
💀
great idea!
@@theAIsearch Great video. Do you know what the specs are to run SVFR and STAR Upscaler mate?
Why are there so many gooners in the gen ai scene 😭
That manga colourisation AI is so cool.
yes!
so many classic manga will be remastered and colored in future, cant wait
@@NostalgicMem0ries Yap. Would make the process of colouring a sequence of images faster.
5:43 Every artist will be overjoyed to get their hands on this art colorizing Ai software. 10:06 That's the way it should be, very realistic and believable, the voices quality is on point. 16:47 I wish that it had an AI to create task outside of Chatgpt platform, and to automate task that has to do with the phone operation itself. 20:29 Now, I'll need to see this robotic performance in real life, it's too good to be true. I don't want to be a skeptic but Unitree robots is breaking some barriers, this marks a defining moment in robotics history. We are getting closing to building a terminator robot, like in the movies. 28:33 This specific feature of geneating a single video from multiple images is mind blowing 😧. Believe me I'm hypnotized 😮 👏🏾👏🏾👏🏾.
The guys who made STAR’s previous models (RealBasicVSR & BasicVSR++) was more advanced in some ways because it grabbed details in both directions of the video, surely they will add that improvement to this STAR model as well.. Wish they would combine these TTA techniques with the video upscaling/super-resolution models that try to identify objects & curves in a video to imply more detail…it feel like both techniques could be used at once. Like if you have a moving object in a low resolution video some frames will not have the object present at all even though it is really still there…this is a huge issue for real world examples like far away cars that have their skinny pillars around the windows disappearing from some frames of the video…or AI models that can only handle downscaled/normalized input videos.
The advancement over other image upscaling technology is that it considers the surrounding frames to grab additional details from. Test Time Augmentation isn’t super new but still useful to know about for other applications.
Thanks for sharing!
@@theAIsearch I have never installed any any model locally , can u tell me how can I install it on my pc?
I have never installed any any model locally , can u tell me how can I install it on my pc?
@@NeverfearGamingg lol that's sort if what this channel does.
good to see all the open sourcing
Perfect no bs intro every time, "AI Never Sleeps". Never change it. Ever.
Thanks!
Ah my weekly dose of AI news ❤
Thanks!
i really love ur work, keep up. its getting really hard to keep updates on AI these days. Especially the good ones.
Perfect… now those early ”cam” releases will soon be watchable 😜
By far one of the best channels for AI news. Thank you for doing the research and bringing the latest news on AI.
you're welcome!
thanks for helping me to keep up!
you're welcome!
5:00 Spot the mistakes, I've spotted 5 wrong color placements so far.
The lateat video upscaling and restoration models are like magic.
thanks for putting all this info together.. very informative! Sub'ed.
Thanks for the sub!
I swear i dont get how some people are still coping about AI being smoke and mirrors. Its insane how quick its going.
ChatGPT tasks might be a game changer for short term investors.
Great video. Do you know what the specs are to run SVFR and STAR Upscaler mate?
Can you look into if there is any news regarding an AI that can help with the in-betweening phase of animation?
I guess it takes ages to scale up my decade old VHS Corn Library.
😃
Congrats on 300k subs!
Thanks!
can you please make video about how to install and use STAR video upscaler?
32:04 This is incorrect. The layers they are referring to are the layers of the model not the "layers" of the scene.
20:39 I don't care how fast they walk, until these things have a proper bend in the foot where the "toes" would be, they're always going to look like they crapped their pants.
@theAIsearch
Is there a demo for RepVideo .. where we can play with it ? (none on Huggingface).
Brilliant, thanks a lot for doing these! Would you have any suggestion on hardware to purchase in 2025 to run these open source models locally?
Thanks. i'd say the 4090 or any cuda gpu with enough vram
I do 2D Animations in anime style and believe me, that Manga Ninja is like a savior for coloring. I know other artist are probably scratching their head causing panic "Oh I'll lose my job." I really hated these type of artist. Unless you do everything on pencil and paper, you have no right to be fucking complaining if you are using digital pen tablet to draw. THESE ARE TOOLS! They supposed to speed up production not replace it. I get that AI generated drawing images are really killing art but that auto color is something already CLIP Studio has but this one is more accurate and beneficial for hundreds of frames for animators. So leave this AI aid for animators. I believe it's much faster to animate and much more not heavy to do when you have auto coloring AI to do that task. Hell in no way I will hire someone to color for me just for extra steps or hand. Unless I'm a big company which I can afford to hire people to pretty much do the entire project for me, hobbyist usually takes months to finish few seconds to minutes of animation. While that's hardwork, this tool really can speed things up and able for artist to simply animate without fearing or being anxious of coloring too much frames later. We can also instead of reusing frames, freely draw as we please since we are now less obligated to color
I think for AI step, this is a plus for artist and animators. These so called Artist AI haters, they just hated everything. I am an artist and I love AI tech and where it is going. It's not like we can stop it anyways so these peeps who oppose it are really wasting time. Use it as a tool. If you really hate it so much, do better. To me personally, this auto coloring is a gift.
thanks for sharing!
It's not tool if do all job, it's tool and user at one😅
youre one of rare artists that doesnt act childish when they hear the word AI. I'm an artist too, and SD really helped me when I couldn't find references to draw, now I'm never having a shortage of refs. I plan to use this too, looks awesome of a tool.
the best (and only) ai tool so far. that will inspire me for my comic even more. i love coloring, but not if you need to paint over and over again for your series.
@2MilyonTM I know right. I know some artist are really deep in to their "do everything by hand." Yeah, no... I want to be able to create more comic series or animationor stories in my lifetime. Doing everything by hand will only get me few series before I get old and get bored and eventually died of old age. With AI, you can produce more, be creative, and focus on just content than worrying how are you gonna multitask your life and your hobby. It's just too stupid to me to see these so called artist opposing a tool. If they truly feel like they are being replace then they are not confident enough with their craft. Also another thing, the stealing part. It's like me "stealing" inspiration from previous artist before me and I have to learn their ways. AI is learning. They don't produce the exact image, they learn from the image and incorporate that learning model into various style. For what purpose, to speed the workflow. I am for speeding the production. It's like you trying to draw Naruto. To them, you stole that drawing when in fact you simply draw Naruto in your own style and inspired.
I do oppose people who prompt AI and claim it as theirs to deceived people or have ai fabricate a speedpaint to fool people. Yeah those ones are something and it kinda creates this paranoia among artist community. As an artist, yeah I don't agree with that but to use it as tool to improve lineart, to do coloring, those are different. You still do art but you kinda let those mundane work to polish it via AI saving you hours and hours of work. You focus on the lines instead that present it and had some creative dynamic knowing you cut off the hardest repeating task.
역시 최고의 AI뉴스 채널이에어요. 좋은 정보 감사합니다. :)
Thank you!
I keep waiting for good AI video special effects to be released.
time to read color manga soon😍
I wonder if Sana will come to comfyui and civitai, or if it's just not compatible. guess time will tell. Not sure if i am gonna try running it locally or not to try it out, if they use their own custom interface then i doubt it'll be anywhere near up to par with features like comfyui.
So now I can watch tron at 4k.
Any AI to make fps higher too? 😁
I know I can use lossless scaling but I want the video
I like your vids because you do the work of "sniffing out" every new development in the world of AI, generative and otherwise. There's so much going on with AI at this point that it's easy to miss something interesting, some new model or new technology. So whenever I see a new vid of yours I must watch it because 9 out of 10 times I'll find out about something new I haven't heard of and that I then wanna try out :D
I dislike your vids because they feel a bit unnecessarily drawn out, so usually I play them at 2x speed, and in the worst cases I just skip chunks. I guess it has to do with the algorithm, the vid must be lengthy probably so there's space for ads, because it's the majority of creators that do it. Like, it's fine, you don't need to handhold me through every single example available in a new project's website and tell me what I'm seeing :D
Same, there's too much going on, it's good to have someone who does the work to centralize all such news in one place.
@NCLDMR I genuinely think this is the most complete and comprehensive channel when it comes to new developments in the AI field, and especially when it comes to new software and new models, compared to any other of "the big ones" by far. It's the one I usually recommend.
Also it's a huge plus for me that it hasn't chosen the way of annoying clickbait titles that many other channels do. I mean titles like "GPT O1 TRIED TO ESCAPE!!1!", and then it turns out that what actually happened is "yeah, so we asked O1 to create a plan of what it would do in a hypothetical scenario where it was a malicious AI that wanted to escape, and it made a plan as asked.."
@ If you do find any other channel that covers cool new AI stuffsthat youd also recommend please let me know. Currently Im only following this channel and the "Mattvid pro AI" one but it's not as comprehensive and complete like this AI Search channel but it does cover some stuffs that this one miss. I probably have FOMO (fear of missing out) with all the cool new weekly stuffs in the AI space LOL
Minimax-01 is really good. I asked to to be a developmental Editor for my novel, 400 pages. Worked really well
what I miss in your videos is showing examples of how the application works by you
thanks for your feedback!
tested SVFR. It's not usable at this time. It's just too slow when compared with codeformer etc. and the quality is probably not that much better, maybe a little.
thanks for sharing!
thank you very much for the Update i really like i check your videos every day to see if there a new Ai update and if you have time can you do a tutorial on the manganinja please.
you're welcome!
19:22 You can NOT run this locally LOL its a 456B model, no single GPU is gonna be able to handle that, i think maybe a single T4 gpu might be able to run a token every second...
Kokoro will be amazing if I can input the voice that I want it to use to generate that TTS, so far it only works with predefined voices from my understanding
6:02 Is That Ash Ketchum from Pokémon?! Is that real?!? 😮
Zuckerberg walks like that
Comparing STAR to Topaz would have been really helpful...
is star upscale any better than topaz?
I see many useful tools especially the text to voice
Why dont ppl combine a lot of AI video and image generators? the quality would be insane
Idk they want them to be independent from each other
Maxmini-01 is also uncensored so you can upload a pick and ask who will die
🤯
There is this image with a guy at e about to push a bolder down a slope. There are people at a b c and d. When you ask most AI who will die it tell you it can't help but this one tells you who will die
Wat was most impressive about the deblurrer was that it left the background soft rather than also sharpening it.
I’m using Windows 10. I used git gui for windows. Installed minaconda and used the gui for python installation. Where I am now is the last command using sudo. Now I’m stuck. Any suggestions? Or is it not for windows 10?
i literally say to the video F5-TTS at the same time as the video XD , that TTS is a monster. Really impressive.
imagine larger scale vidu2 style references to video, with detailed ai storyboarding, with veo2 quality.
No one will make movies with this technology anyway, except for AI enthusiasts. Camera, action, done. Nolan will continue making his movies and making money.
@ yeah but for people who want to make fanmade stuff, or people who want to bring their ideas to life without access to actors, fancy equipment and stuff. Just for fun. But anime studios i think wont make complete ai stuff, but will use this as a tool. I could see the ai coloring being a big help, although yeah many anime creators would disagree with it
@@prodvxmp This is what anime is famous for: the love for detail in every frame. In order for people to accept it, we need to spend hours for ten seconds, but RUclips is filled with lazy clowns throwing shade at AI.
@@TheCurseOfBlago yeah but some studios are stressed af and coloring only some frames while autocoloring the rest could be potentially helpful if u only wanna color every two frames or so. Maybe for low budget ones who knows.
@@prodvxmp I don't argue, automatic coloring looks fun and useful.
I wish I knew how to follow instructions so I could install MangaNinja.
First robot to run? Boston dynamic did it first
Those robots are actually so cool, I can’t even tell if the videos of them are CGI or not but I hope they aren’t haha
Its not CGI or AI generated. Its real, the company making it has a youtube channel posting their progress over the years
LOVE FROM PUNE MAHARASHTRA🇮🇳🇮🇳🇮🇳 😎🥰😘❤👍🧡😊🤝😍❤😎
Yup !,same
36:39 I just want to ask a genuine question, if the language model uses both memory retention and guessing technique to solve problems, but the guessing technique is what dominates at the end, how then will y'all achieve super intelligent with a guessing method. For example, if I'm preparing for an EXAM, and I will need to memorize for it, wouldn't I need to depend only on my memory to pass the exam or test? I can't be guessing to pass that examine or test. I'm struggling to understand how we'll achieve ASI by using a guessing approach AI model. 😎💯💪🏾👍🏾 I'm not bashing here, I'm just asking a genuine question.
Can you do a tutorial on how to install star. cant get it to work, I'm running a windows PC
after watching all this i think ineed to upgrade my sytem
Hailuo AI is the best one for creating stories IMO. You want an idea for viral video Hailuo is where you go.
Can you test hallo3? No one is talking about it…
Sana is Apache 2.0 now?!?!
😃😃😃
21:01 "On your right!"
Freaking insane stuffs!
I would like to know when is x portrait 2 coming by ByteDancs
And why do we need AI to make the video clearer? Pictures are made clearer without AI, and videos are a queue of pictures.
There are some areas where it is impossible to use another method. For example, to find something similar to something.
Big ai, big video 🦾🦾
You could color manga quickly whit this Manga Ninja.
yes!
I have never installed any any model locally , can u tell me how can I install it on my pc?
Gracias
Sir, no more ai model training?
The upscaling still looks cartoony, like a filter has been added. AI is still struggling with adding detail that doesn't exist. And this is a good thing.
*for now
"Like this" "This is what you get"
There are so many classic corn that should be 4x sharper.
great idea!
Can someone tell me the best image upscaling website or tool
Hold on now. Minimax is as good/better than GPT4o and Claude, is free, and can be run locally?
Just another normal day or did I miss a caveat?
it's better at handling large contexts, like if your prompt is super long
Thanks. By the way there's no such thing as nine point eleven. After a decimal place you read rhe numbers out separately so 9.11 is nine point one one, not nine point eleven (which is incorrect)
That rule is there because it massively reduces the number of actual humans that get confused!
China is leading the AI space
Nope. They’re mostly leading the propaganda space 😂
India
Nah, even the Deepseek CEO says China is just tagging along for the ride and copying stuff... but he wants to change that and make them an active contributor to frontier innovation.
@isitanos this is a case where the copy is better than the original because if I can ask a model anything and not see the dreaded I can't help you with that then it's better especially since I didn't even jailbreak
How many times can they use the word super
OpenAI shill channels flew over MiniMax like it was nothing, but it's actually great 🙂
The comment section always proves how everyone is not ready
wym?
@marcoplankensteiner plenty of people think alot of robot videos are fake
@@camronrubin8599 for real, they think its AI generated.
But I get why they would react like this, the progress is just too fast for us to believe.
Just a few months ago, robots would walk like they had stroke every half a second
MangaNinja gonna help make anime faster GOOD more fantasy Anime for me
sounds like Repvideo is extremely restrictive
21:02 is this video not AI generated???? lmao look at the guy in blue in the background this cant be a real demo
The person in blue is a woman and shes just running weird? Thats not indication its ai generated video. Use a more critical eye, no AI video generator can maintain that level of consistency with characters, they would all looks slightly different in each video, different ports, different proportions, different colors...etc.
That being said, there is a possibility that this is an motioned tracked comp shot which means they have a real 3D model that is rigged, and they replace the person with it. But that still does not make enough sense as the people reacting to the "robot" is too weird for it to just have been a normal person standing there.
Its real. They have a bunch of demos over the years. And if you really want proof that its not AI generated, you can analyze the chinese texts and see that theyre accurate in the video. AI cant do that. (yet).
Death of creativity on it's way
Nah, if anything it's gonna get a bopst in value with all this ai slop coming out
👍✨
20:25 Those are all AI animation. Is that what you meant?
Its not AI animation
ChatGPT Tasks + Stocks + Google sceen record = !!!!!!!!!!!
TTS yes!!!!
Zoom in… Enhance
Modelling is pretty much finished now. I wouldn't say it's a big loss to society as its contribution wasn't high due to not needing any skill at all
32:25 better than Hulan or Machi ??? can you please write the 2 names of video gen you said here please?
hunyuan & mochi
ai is awesome its one of the reasons i have hope lol
👋 hi
hi
Balls teehee
I got the first Ai toilet paper
Zoom in and Enhance about to be real
Using AI to upscale AL. It's AI-ception!
If physics are anything to go on, the majority of all the super impressive robot and drone swarms on youtube are fake.
I want to generate games not videos
Bro just pay a photographer lmao those portfolios look fake as hell
Colorists, start looking for new career!
Seriously