lemme guess, another 3d model generation ai tool that looks impressive, but it's not publicly available and all we got is a research paper that no one will be bothered to read, this is getting old
There is a few open source ones he's mentioned that are available though. You just need a computer to run it as no one that works on these things for free on there own time is going to pay to run a server for other people to use there AI on there machines, that is way to costly.
as an animator I actually think AniDoc is what a lot of animators have been looking for. we want to do the fun creative process, not the hard time consuming parts.
@@Ristaak Witness testimony is and will always be unreliable. Mainly because people can't be relied on to tell the truth but also because our memory is not as accurate as we like to think it is.
3:03 I like that they added the ability to draw a box around the part that you want to edit. I just wish it was possible to also make a selection with a brush tool, so that we can make a more precise selection on the part that we want to edit. I think precise selections would be important for more complex 3D objects.
I always look forward to your videos, I love the style you present the information! Also I feel that I should say, I appreciate the way you add many different AI's for different uses. I understand that some people are only interested in BIG AI news like video generation or LLM models etc. Just know, there are people like me that appreciate the more obscure or less explored scenarios for AI such as medical, scientific, design etc. So I hope you continue adding this type of information as well. I feel that it really sets you apart from every other AI channel that covers the same BIG AI news.
Such a great AI news channel. Thank you for your time and energy. That said, maybe a flag or visual or auditory signal at the head of each chapter that denotes whether its available or not instead of waiting till the end. Just a suggestion. You rock!
First all, I love your videos , they are straight to the point and this is becoming more and more important since news are an avalanche nowadays. Robots demo are insane. I'd like to see robots ascending instead of descending. Even though the riding demo part blown me away. The HSM demos are impressive until you realize first demo has 4 person and the 3d representation missed one person. Even though is an outstanding technology, almost SCI tech.
23:21 I can already see human workers on production lines and in offices ordering robots to do backflips. lol Thanks the important news! All of them very awesome! Especially AniDoc is very good news for me as an anime fan!
thanks for video, always nice to see new AI things. As only an enthusiast, i don't see any of these being something i'd play with personally but it's all cool!
Thank you so much AISearch, for educating and keeping us informed with this content! Please don't forget, however, that these robots should stay slow! While we can't say that would stay forever, another plan must be constructed in order to prepare for their fleet! Thank you!
I wonder when the AI comes that can take any comic book that you scan in, then on itself create an entire moving animation out of it in the style of the person who drew the comic, and change the text baloons (or however these white rounded areas with their spoken words are called in english) will be spoken with ai voices. So basically generating an X-men animated kind of " tv series" out of X-men comics (just say X-men as it has a well known tv series and comics)
Hmm, technically, that would be like combining about 8 or more AI tools that already exist, so it's possible to do, the generation time might take a while, and the voices the AI uses probably wouldn't match a voice actor you might choose, but generally there are particular types of voices for particular types of characters, and if the AI is trained on a lot of animated content and voice types, it could choose the appropriate voice, just like music generating AI always puts an appropriate voice for a particular type of song.
23:50 All those demos are mostly irrelevant, the company that will have the most data out of there robot will learn to do useful stuff. The name of the game is who gonna make the most robot the fastest.
Well, Michael Reeves already taught a Spot robot to pee beer, so the potty training will come in handy, one future feature that I think future robot androids will have is the ability to drink a liquid and filter and cool the liquid to be pure water, technically, and to make the android anatomically correct, they will probably make them so the android pees the water, alone with the android in a hot desert and you forgot to bring water? No problem, just give the android some of your urine in some container for it to drink and you'll have fresh cold water soon. Yeah, it's probably a better feature if you're not aware of the details of how the android is making the water.
i decided to make anthro art and now im going for photorealistic anthro since its scarce, next i guessil go to 3d so facelift and other modeling ai could be be very helpful at that point
Let robots do acrobatic stuff is actually quite easy, using predefined motions - it's basically the same as using a motion capture data file for a rigged 3d figure. It's way harder, however, to let a robot do this by ITSELF, so I think that most companies are actually faking this with mocap techniques to entertain their investors.
There is AI software and algorithms that can do 3d motion from watching a video of a person, so if you add this to a robot than all you have to do is let the robot see a video of someone dancing and then it can use the 3d motion to direct it's movements. Or if you can dance, you could dance for the robot to see.
@@onidaaitsubasa4177 It's not only about "knowing" the motion - the robot would need to take its weight, friction, motor forces and lots of other physical forces into account, but's this is simple Newton physics, which can all be calculated.
I saw a video from China where Unitree failed noticeably with their robots which were constantly falling, losing direction and needed a support of to people, so I wouldn't jump to any conclusions, although the demo in this video looks impressive.
Hopefully that was early video footage, with added training comes better movement and balance, so it can still be improved from that, I remember a video that showed their latest bipedal robot can walk much better than the earlier version, it's a lot like when a child learns to walk, they're not too good at first but added practice and soon they're walking really well.
14:35 i don't wanna jump to conclusions, but imagine it's literally just bc it's a chinese model 22:35 bro just dropped a bomb and then carried on like nothing happened lmfao
CAN YOU PLEASE PUT A RED OR GREEN SPOT IN TE CORNER OF PARTS OF YOUR VIDEOS SO WE DO NOT HAVE TO WASTE TIME LISTENING TO AI TOOLS WE CAN NOT EVEN USE AND CAN DIRECTLY SEE WHAT WE CAN USE OR NOT. GREEN DOT = USABLE/DEMO RED DOT = NOT USABLE ATM. Or just make 2 sections in your videos where one is about the tech we can see but not use and hte other is stuff we can use. It's a bit like going to a mansion that is for sale but that you'll can not even afford. It feels like a waste of time to inspect it and check if the rooms would fit your stuff if you can't even buy it.
You should've did a step by step tutorial. I never get things to work when installations are too complex. A previous tutorial had me require getting a new PC.... :/ We know trellis, but some of us don't know what micromamba even means...
I don't think it matters if it's open source if your graphics card isn't meant for AMD use cases. I have an eye for the fact that I have an AMD card because I don't want to pay an expensive price. I just wish they could make it publicly available online so that we don't have to require any additional setup.
It's probably doing so at the moment in small ways, but that may become a larger impact in the near future. The fact it's already at the point that they are using AI programs to generate data to train more AI with some success is wild.
I am interested in an AI model that will allow me to describe my trading technique and get it to auto trade! My technique is different to most traders! Such a model may also glean information from my trading charts visually, to assist in building a trading system! Information can be verbal, script or image based, or a combination of all three! ! Do you know of any such models out there? So far I have a blank on this one, but surely it has to be possible?
Had to stop watching when you made it clear you don't understand areas of focus in robotics. Some bipeds have high fine motor skills-sorts of dexterity, with acrobatics not being an important consideration at the moment. Note that neither Unitree nor Boston Dynamics have developed highly capable hands. It would seem thia channel is for people who don't usually pay much attention to these topica.
Some of those 3d model in the 'articulate-anything' demo are existing human made and are retrieved and animated, no way AI generated 3d models can be that good right now.
lemme guess, another 3d model generation ai tool that looks impressive, but it's not publicly available and all we got is a research paper that no one will be bothered to read, this is getting old
yep it sure is..
yeah, should start ignoring this crap without demo
Right on
Some are open source
There is a few open source ones he's mentioned that are available though. You just need a computer to run it as no one that works on these things for free on there own time is going to pay to run a server for other people to use there AI on there machines, that is way to costly.
as an animator I actually think AniDoc is what a lot of animators have been looking for. we want to do the fun creative process, not the hard time consuming parts.
its still gonna be hella inconsistent tho
@qirmi2030 true but it should get better overtime
But for 3D artists it's not good
@@Pxilez3d this doesn't even apply to 3d artists
@@krasen671
Oh😅 i though you were talking about 3D model generator, i didn't look the names
Facelift 3D is going to be very handy for creating LoRA's
oh good point!
Now visual proof are legit fucked, what are the point of courts anymore 😭
over bro
I mean, it'll be back to how it was in the 1800's. Witness testimony is always king.
Real camera with water marks
@@Ristaak Witness testimony is and will always be unreliable. Mainly because people can't be relied on to tell the truth but also because our memory is not as accurate as we like to think it is.
Lawyers make pennies anyway they are ready to quit
3:03 I like that they added the ability to draw a box around the part that you want to edit. I just wish it was possible to also make a selection with a brush tool, so that we can make a more precise selection on the part that we want to edit. I think precise selections would be important for more complex 3D objects.
4:50, that's impressive but still shadows/lighting and deformed drawing need to be colored by hand but this increase the speed a lot for animators
Thanks
Facelift 3D seems like a super exciting tool.
😃
Dude we are going to have all the anime created in a week when this A.I gets good.
Think of all the amazing manga that has not been turned into anime
i am also using it to create anime videos lol
Yooo the ai that colorizes a frame it's crazy that's huge for the animation industry, and for the animators
This channel has become just as important to me as the local news channels.
wow, thanks!
@@theAIsearch you earned it, just keep it up. Somebody has to recon this new age and come back with some usable information. Big responsibility
this is insane stuff, i'm so glad i'm living through this
Not for long time😂 judgment day is comming
@@marianpe5773Did a little birdie tell you that?
Retired newspaper cartoonist in Utah… Have fun… AI is a game for the young artists.
No
Hi,
Would you be open to sponsorships?
22:30 That's always a constant thought in my head lmao 😅
HAPPY NEW YEAR AI Search!!!
I always look forward to your videos, I love the style you present the information! Also I feel that I should say, I appreciate the way you add many different AI's for different uses. I understand that some people are only interested in BIG AI news like video generation or LLM models etc. Just know, there are people like me that appreciate the more obscure or less explored scenarios for AI such as medical, scientific, design etc. So I hope you continue adding this type of information as well. I feel that it really sets you apart from every other AI channel that covers the same BIG AI news.
Thank you so much!
AI search always got us with the latest AI news, much love ❤
Thanks
Such a great AI news channel. Thank you for your time and energy. That said, maybe a flag or visual or auditory signal at the head of each chapter that denotes whether its available or not instead of waiting till the end. Just a suggestion. You rock!
First all, I love your videos , they are straight to the point and this is becoming more and more important since news are an avalanche nowadays.
Robots demo are insane. I'd like to see robots ascending instead of descending. Even though the riding demo part blown me away.
The HSM demos are impressive until you realize first demo has 4 person and the 3d representation missed one person. Even though is an outstanding technology, almost SCI tech.
23:21 I can already see human workers on production lines and in offices ordering robots to do backflips. lol
Thanks the important news! All of them very awesome! Especially AniDoc is very good news for me as an anime fan!
thanks for video, always nice to see new AI things. As only an enthusiast, i don't see any of these being something i'd play with personally but it's all cool!
you're welcome!
DeepSeek2 or 3 is also thinking model like o1 or Gemini flash. There is a button to switch thinking on/off ^^
Amazing video! Keep it up! Love the template and style you give information!
Thanks!
Thanks for sharing as always. And if people are tired of it, no one is forcing anyone to 'consume' any of the content here.
And here I thought I was the one being up to date on ai. Where do you find all this stuff man.
Thank you so much AISearch, for educating and keeping us informed with this content! Please don't forget, however, that these robots should stay slow! While we can't say that would stay forever, another plan must be constructed in order to prepare for their fleet! Thank you!
That last one would be great if they also had a Gaslands LORA.
i cant use anidoc with my own clips and images it gives an error
I wonder when the AI comes that can take any comic book that you scan in, then on itself create an entire moving animation out of it in the style of the person who drew the comic, and change the text baloons (or however these white rounded areas with their spoken words are called in english) will be spoken with ai voices. So basically generating an X-men animated kind of " tv series" out of X-men comics (just say X-men as it has a well known tv series and comics)
Hmm, technically, that would be like combining about 8 or more AI tools that already exist, so it's possible to do, the generation time might take a while, and the voices the AI uses probably wouldn't match a voice actor you might choose, but generally there are particular types of voices for particular types of characters, and if the AI is trained on a lot of animated content and voice types, it could choose the appropriate voice, just like music generating AI always puts an appropriate voice for a particular type of song.
Ai, progress is so damn fast, can't even play around with a tool, then new tool came😂
yes!
I never saw notification of when a model presented here had got the open source project to install....
23:50 All those demos are mostly irrelevant, the company that will have the most data out of there robot will learn to do useful stuff. The name of the game is who gonna make the most robot the fastest.
Which LLM works better for coding with 12 GB of VRAM? Deepseek V3 or Qwen 2.5 Coder? Or something completely different? 🤔
The FaceLift looks good.👍🏼
yes! hope they release the code soon
We've reached the point where we can potty train robots.
Well, Michael Reeves already taught a Spot robot to pee beer, so the potty training will come in handy, one future feature that I think future robot androids will have is the ability to drink a liquid and filter and cool the liquid to be pure water, technically, and to make the android anatomically correct, they will probably make them so the android pees the water, alone with the android in a hot desert and you forgot to bring water? No problem, just give the android some of your urine in some container for it to drink and you'll have fresh cold water soon. Yeah, it's probably a better feature if you're not aware of the details of how the android is making the water.
I want to try out DeepSeek
i decided to make anthro art and now im going for photorealistic anthro since its scarce, next i guessil go to 3d so facelift and other modeling ai could be be very helpful at that point
Let robots do acrobatic stuff is actually quite easy, using predefined motions - it's basically the same as using a motion capture data file for a rigged 3d figure. It's way harder, however, to let a robot do this by ITSELF, so I think that most companies are actually faking this with mocap techniques to entertain their investors.
There is AI software and algorithms that can do 3d motion from watching a video of a person, so if you add this to a robot than all you have to do is let the robot see a video of someone dancing and then it can use the 3d motion to direct it's movements. Or if you can dance, you could dance for the robot to see.
@@onidaaitsubasa4177 It's not only about "knowing" the motion - the robot would need to take its weight, friction, motor forces and lots of other physical forces into account, but's this is simple Newton physics, which can all be calculated.
I saw a video from China where Unitree failed noticeably with their robots which were constantly falling, losing direction and needed a support of to people, so I wouldn't jump to any conclusions, although the demo in this video looks impressive.
Hopefully that was early video footage, with added training comes better movement and balance, so it can still be improved from that, I remember a video that showed their latest bipedal robot can walk much better than the earlier version, it's a lot like when a child learns to walk, they're not too good at first but added practice and soon they're walking really well.
23:10 the Unitree G1 looks to be computer animated
i agree
no it doesn't
22:09 meet our future police force. No joke. This is scary stuff for keeping control of the general public.
imagine someone hacking this and therefore controlling the entire 'police force' of an area
@@theAIsearch Hmm, and then using them to facilitate true freedom. It's a promising concept 🤗
22:30
yes, that's what i thought first
The coloring has many defects, I would hate to notice these in an anime
Every other week, a new AI program 1ups the previous model.
14:35 i don't wanna jump to conclusions, but imagine it's literally just bc it's a chinese model
22:35 bro just dropped a bomb and then carried on like nothing happened lmfao
Hey, if you're looking for new content ideas, maybe you could help me with the animate-x installation because I found so many errors when doing it.
CAN YOU PLEASE PUT A RED OR GREEN SPOT IN TE CORNER OF PARTS OF YOUR VIDEOS SO WE DO NOT HAVE TO WASTE TIME LISTENING TO AI TOOLS WE CAN NOT EVEN USE AND CAN DIRECTLY SEE WHAT WE CAN USE OR NOT. GREEN DOT = USABLE/DEMO RED DOT = NOT USABLE ATM. Or just make 2 sections in your videos where one is about the tech we can see but not use and hte other is stuff we can use.
It's a bit like going to a mansion that is for sale but that you'll can not even afford. It feels like a waste of time to inspect it and check if the rooms would fit your stuff if you can't even buy it.
CANT WAIT ANIDOC 2.0😊😊😊😊
😃
now we need an AI that can generate that lineart video.
@@spraygospel5539 no
guess ill go back on making animated video then. AniDoc might help me with my work.
😃
Can you make some test about phind, nestmind ai studio and qwen
any other good vpoice changers??
Are all these AI tools programmed from zero? Or do they use ready libraries? How does it happen?
most are built upon existing libraries & datasets
So why are you using already existing FILMS to demo the AniDoc. How are we sure those are not real frames from the film.
WE want to see something NEW
AI never sleeps, but does it take time off for holidays?
You should've did a step by step tutorial. I never get things to work when installations are too complex. A previous tutorial had me require getting a new PC.... :/ We know trellis, but some of us don't know what micromamba even means...
Give a solution where can make free Motion Graphics to illustrate financials videos
amazing AI news... thank you!
no problem!
Ai here we see all
Imagine training DrivAerNet and forgetting to include rims data...
Impressive indeed 👌🏻
😃
Open Weights*** Not open source (unless they also released all the data and code used to train the model)
i asked ‘what letters repeat themselves in this sentence “and it got it wrong
weak week, tvm for the video
Robot dog on wheels idk but its funny😅
With 1 bit quant deepseek is only 20gb. 🥺
23:01 has been seen walking like a grandpa. th-cam.com/video/12IwfzyHi0A/w-d-xo.htmlfeature=shared
This 3d models maybe look nice but when put in blender is mess up 😅
The chair frame was low quality
>knows who anya is
>knows what jjk is
>knows what demon slayer is
why hello fellow weeb!
😃😃😃
LOVE FROM PUNE MAHARASHTRA🇮🇳🇮🇳🇮🇳 ❤🤝🧡🥰😉👍😀😘😎😃😍😊🥰❤
Thanks!
I need some realistic head
truly sickening! my guy doesnt get any time off.. haha
Great video as usual
I cannot believe, I just completed the whole video in 3 minutes
You usinG Ai?
Fact check the unitree videos before posting it.
Lets goo deepseek!
is anidoc free and open source?
yes
OpenAI is behind
Thank you
I don't think it matters if it's open source if your graphics card isn't meant for AMD use cases. I have an eye for the fact that I have an AMD card because I don't want to pay an expensive price. I just wish they could make it publicly available online so that we don't have to require any additional setup.
??? wdym "isn't mean for AMD use cases"?
@@arcadesmasher majority of these AI tools don't work for AMD Users
@@NoahtheGameplayer ohh. yea i get that
Let's gooooooo 🗣️🗣️🗣️
😃😃😃
dont sponsor abacus ai, they have jittery scroll and no customer care
Is it true that A.I. helps A.I. make better A.I.? 🐿️
O1 is currently doing that
It's probably doing so at the moment in small ways, but that may become a larger impact in the near future. The fact it's already at the point that they are using AI programs to generate data to train more AI with some success is wild.
squirrel guy 🐿
th-cam.com/video/2Usnz0MdNbc/w-d-xo.html
I am interested in an AI model that will allow me to describe my trading technique and get it to auto trade! My technique is different to most traders! Such a model may also glean information from my trading charts visually, to assist in building a trading system! Information can be verbal, script or image based, or a combination of all three! ! Do you know of any such models out there? So far I have a blank on this one, but surely it has to be possible?
is so insane And if in 2025 it will Got ai Smarter and more capable than in 2024 ☠️
open-source is the best
When can I get a fem-bot?
Had to stop watching when you made it clear you don't understand areas of focus in robotics. Some bipeds have high fine motor skills-sorts of dexterity, with acrobatics not being an important consideration at the moment. Note that neither Unitree nor Boston Dynamics have developed highly capable hands.
It would seem thia channel is for people who don't usually pay much attention to these topica.
How long until you admit that YOU'VE BEEN AN A.I. THIS WHOLE TIME!! @AI Search
As a large language model, I'm unable to comment on the above
That AI car design thing is a bunch of BS. What is going to happen is any new car comes out they will claim they designed it and then demand payment.
the non-commercial licence is indeed very restrictive
22:35 😳😳😳
CCP Deepseek is pleased
nice try
Some of those 3d model in the 'articulate-anything' demo are existing human made and are retrieved and animated, no way AI generated 3d models can be that good right now.
Niceeeeee
😃
i need more anime then i can watch hope the can make dub same way using ai
1st
😃😃😃
LET'S BUILD THAT WALL Y'ALL ONE BRICK AT A TIME! 🧱
ⓘ 𝘠𝘰𝘶 𝘤𝘢𝘯'𝘵 𝘴𝘦𝘦 𝘵𝘩𝘪𝘴 𝘤𝘰𝘮𝘮𝘦𝘯𝘵 𝘣𝘦𝘤𝘢𝘶𝘴𝘦 𝘺𝘰𝘶 𝘢𝘳𝘦 𝘶𝘴𝘪𝘯𝘨 𝘢 𝘊𝘩𝘪𝘯𝘦𝘴𝘦 𝘮𝘰𝘣𝘪𝘭𝘦
1st
😃😃😃