AI turns video into layers, makes 4D videos, new open-source video, SORA leak, realistic 3D models

Поділитися
Вставка
  • Опубліковано 3 гру 2024

КОМЕНТАРІ • 168

  • @theAIsearch
    @theAIsearch  3 дні тому +6

    Thanks to our sponsor Abacus AI. Try their new ChatLLM platform here: chatllm.abacus.ai/?token=aisearch

    • @dragonifyamazing2721
      @dragonifyamazing2721 3 дні тому

      doesn't 28:37 sora leak is only a turbo model. I am pretty sure their big one are still hidden from the public. I swear those leakers are salty they are not getting paid

    • @El__AnTiGuO__cArIFlO
      @El__AnTiGuO__cArIFlO 3 дні тому

      @@dragonifyamazing2721 agreed

  • @1conscience0dimension
    @1conscience0dimension 3 дні тому +13

    What impresses me the most is how quickly humans are coding all of this, yet I can barely keep up with even average coding. But that's a different matter. In truth, I often wonder what I am truly capable of. External goals seem fleeting and illusory. The only real goal is to be at peace within oneself, to simply be. As the world speeds ahead, with machines increasingly controlling what we once did, the internal pursuit remains constant: to be present, aware, and true to oneself.

    • @SrFrederick
      @SrFrederick 3 дні тому

      Artificial Intelligence was the key to gaining a x5 boost to develop faster (which makes up for the poor quality)

  • @xandragonist
    @xandragonist 3 дні тому +35

    green screen man is no longer needed

    • @kreekkrap
      @kreekkrap 3 дні тому +1

      AI taking everyone's job before 2026 💀

    • @1conscience0dimension
      @1conscience0dimension 3 дні тому +7

      You don't see, but it's now the invisible man

    • @El__AnTiGuO__cArIFlO
      @El__AnTiGuO__cArIFlO 3 дні тому +1

      @@1conscience0dimension makes sense if we put into example

    • @lamuel06
      @lamuel06 2 дні тому

      That depends but mostly yes.

    • @kreekkrap
      @kreekkrap 2 дні тому +1

      @@xandragonist That's depressing.

  • @ianwatson5767
    @ianwatson5767 3 дні тому +8

    Omni control looks insane! This is the future marketing visualization.

  • @vrshock8790
    @vrshock8790 3 дні тому +9

    XR and MR are on the horizon for something big, I mean something really big.

    • @mostlymento3068
      @mostlymento3068 3 дні тому +1

      Im so pumped

    • @hiddendrifts
      @hiddendrifts 3 дні тому +1

      tbh i've always thought that ar and adjacent technologies would have significant applications in the entertainment industry

  • @crippsverse
    @crippsverse 3 дні тому +5

    I love that you never hype things and always manage to have at least one (often more) unique items to report on. My favourite AI news channel.

  • @NeuralDev
    @NeuralDev 3 дні тому +3

    the speed of those release is absolutely insane

  • @armondtanz
    @armondtanz 3 дні тому +11

    Lol. This channel should come with health warning.
    Literally mind blowing.

    • @Mubashar783
      @Mubashar783 3 дні тому +2

      Ai never stoping making me 😮

    • @theAIsearch
      @theAIsearch  3 дні тому +4

      lol

    • @BoyFromNyYT
      @BoyFromNyYT 3 дні тому +1

      @@theAIsearch hey do u know any tutorials about how (In general installing all these open source things locally... like the github and stuff cause its confusing and i dont know where to start... also are you still planning on making the frame interpolation tutorial??? or is there one i can watch to run that.

    • @minh3503
      @minh3503 3 дні тому

      haha

    • @Mubashar783
      @Mubashar783 3 дні тому

      @@BoyFromNyYT He won't reply

  • @DefinitlyAPerson
    @DefinitlyAPerson 3 дні тому +6

    Video stabalization is going to be crazy.

  • @276-
    @276- 3 дні тому +4

    you should brand the ai news similar to an actual news channel's style, i love these videos man :)

  • @silverony8305
    @silverony8305 День тому

    Generative Omnimatte is by far the most mind-blowing AI tool I have ever seen in a while.

  • @Ali_Alhakeem
    @Ali_Alhakeem 3 дні тому +2

    These ai are actually really helpful and can help productivity alot.
    really impressive jump .

  • @tatecheddar
    @tatecheddar 2 дні тому +1

    Now take Samurai and implement it into a control grid with access to any public or CCTV camera. SCARY!!

  • @mandyandy-e8x
    @mandyandy-e8x 2 дні тому +2

    Think about how many jobs this will replace, OVERNIGHT!

  • @yikifooler
    @yikifooler 3 дні тому +3

    Whenever someone says the word "Consistent" in a AI field..... I laughs and says what a joke.

  • @Retrogamer71
    @Retrogamer71 3 дні тому +1

    Video production jobs in advertising are going to be available only to a couple of digital specialists.

  • @MichaelChanslor
    @MichaelChanslor 3 дні тому +2

    Thanks for all your work with these videos!

  • @RasmusSchultz
    @RasmusSchultz 3 дні тому +2

    32:28 the irony of billing Claude as a "non-thinking model", when it was literally the first to introduce the tags before o1 was even a thing 😏

  • @dhillaz
    @dhillaz 3 дні тому

    The edge highlighting at 6:00 is my favorite use case. While other features are 'cool' this would really improve instructional videos.

  • @ThomasJDavis
    @ThomasJDavis 3 дні тому +1

    So basically:
    Generative images = visual coherence
    Generative image sequences/videos = visual coherence across multiple generated images
    Generative 4D videos = videos/image sequences with depth coherence
    Generative gaming = videos/image sequences either with or without depth coherence that also accounts for user input in real time.

  • @hiddendrifts
    @hiddendrifts 3 дні тому +1

    27:02 part of me thinks this was some sort of "false flag" operation to make artists look bad. bc as you say, the program was entirely voluntary. if you didn't agree with their terms and conditions, you didn't have to participate. the only credit i'd give them was maybe if sora was claimed to be made "in partnership with leading artists" but instead just wanted to use them as pr, but otherwise i don't see the problem
    31:34 that is most definitely a "backronym" lmfao

  • @zengrath
    @zengrath 3 дні тому +1

    I don't understand the Samurai one, video and site shows how it can detect and keep track of an object, but not what you can do with it.

    • @spookyscarygraviton5944
      @spookyscarygraviton5944 3 дні тому +2

      Video editing. When you need to apply effects or lock camera on a specific object or character

  • @mickelodiansurname9578
    @mickelodiansurname9578 День тому

    I've worked in IT for 40 years and over the last few years AI and ML - and its getting Sci Fi, and its getting Sci Fi fast!

  • @ZorenStudio55
    @ZorenStudio55 3 дні тому +4

    Ai never sleeps 😅

  • @manonamission2000
    @manonamission2000 3 дні тому +2

    simulated time-travel might not be that far off

  • @GarageDayzNite
    @GarageDayzNite День тому

    when you have time, could you make an installation / tutorial video for Samurai object segmentation?

  • @williamdoust
    @williamdoust 2 дні тому

    Perhaps LTX video could be used for background video loops if abstract textures

  • @javaboy6581
    @javaboy6581 2 дні тому

    Increíbles tus vídeos, deberían de ser traducidos por google a todos los idiomas automaticamente

  • @orac229
    @orac229 3 дні тому

    Any recommendations for AI that could extract car number plates from video. I have Dahua ANPR cameras but sometimes it is not clear but frame by frame you can get different numbers of some are blurry but as a person I can guess the likely number but the camera can not, (and other issues with exposuer or speed blur or other issues) so is there any AI that might be helpful or being developed?

  • @MrNobodyX3
    @MrNobodyX3 4 години тому

    8:01 dude, you should see magnetic mask on Final Cut 11 it's basically this

  • @PovertyHelping
    @PovertyHelping 3 дні тому

    Thanks so much for this opportunity. This is the best online classes. 🏭
    Perfect for gaming.

  • @hogarthheathan
    @hogarthheathan 3 дні тому

    What do you think is the best llm to download, where can you check the bench marks?

  • @TechnoEdit500
    @TechnoEdit500 3 дні тому +3

    brooooo what is happening here are we living in 2024?

  • @DataJuggler
    @DataJuggler 3 дні тому

    Texturing PBR''s is really cool. Often if you export to PBR textures are lost.

  • @NeoRelic-o8p
    @NeoRelic-o8p 3 дні тому +2

    3d modeling!!! Let's go!!! 🥳🎉

  • @iaincampbell4422
    @iaincampbell4422 3 дні тому

    Testing omnicontrol repeatedly on the hugging space workspace from the link and it seems to completely ignore the reference image I upload??? Not working, am I missing something???

  • @jeanvirginio3064
    @jeanvirginio3064 3 дні тому +1

    Thank you for the video!

  • @dreamzdziner8484
    @dreamzdziner8484 3 дні тому

    Oh my AI! Every single day something mind blowing comes out.

  • @testales
    @testales 3 дні тому

    A 32b model beating Claude Sonnet in coding with 50:36, these benchmarks are getting more useless and absurd every day. But these 4D models are very interesting because other than just jiggling around with the view, it would allow for 3D videos I guess! Even converting regular videos into VR videos might be possible in principle.

  • @randomcommenterurl
    @randomcommenterurl 3 дні тому

    CAT4D type technology would be great for no crop video stabilization

  • @gersoncar1472
    @gersoncar1472 2 дні тому

    Great video as ALWAYS! could you do a tutorial on how to use Samurai?

  • @juiceman110
    @juiceman110 3 дні тому

    Assuming basically linear scaling laws, if this Qwen model was the size of ChatGPT 4o, then it would for sure be AGI!

  • @S.I.F.A.R
    @S.I.F.A.R 3 дні тому +1

    we need this on adobe rotoscoping

  • @NeverfearGamingg
    @NeverfearGamingg 3 дні тому

    Gonna use SAMURAI but..will it work on AMD RX 6750Xt?

  • @ShA-ib1em
    @ShA-ib1em 3 дні тому +1

    9:28 Itzy spotted.. video clicked.

  • @GeorgeDimovelis
    @GeorgeDimovelis 3 дні тому

    why is there big progress on ai text to video but text to img hasnt got anything new recently

  • @apatsa_basiteni
    @apatsa_basiteni 3 дні тому

    Damn all these are crazy good loved the Omnimate

  • @fitybux4664
    @fitybux4664 3 дні тому +1

    15:50 Ok, professional clothing models are DONE! 😆 "Put this new clothing design we made onto a man/woman." And... done, put it in your online store and sell. 😆

    • @theAIsearch
      @theAIsearch  3 дні тому +1

      yes, it's so easy now!

    • @nicosoftnt
      @nicosoftnt 3 дні тому

      ​@@theAIsearchYou make it sound like it's a good thing for us, we'll potentially need UBI within 5-10 years or there will be trouble

    • @iaincampbell4422
      @iaincampbell4422 3 дні тому

      Yes, except I tried omnicontrol's hugging space and it seems to completely ignore the image I upload and reference as this item in the prompt. Not sure if I'm missing something!

  • @28Corner
    @28Corner 3 дні тому

    Hey sir, have you tried the Meta Demo Lab new AI? It would be interesting if you make a video about it

  • @ivansoto9723
    @ivansoto9723 3 дні тому +1

    Ratatouille over here making a pizza lmao. Cute.

  • @Lv7-L30N
    @Lv7-L30N 2 дні тому

    gracias mi amigo tu siempre traes los mejores avances

  • @pallenda
    @pallenda 3 дні тому

    27:48 no snow falls of the pine tree.

  • @aa-xn5hc
    @aa-xn5hc День тому

    Upfront say whether usable now or not each is

  • @g4u553
    @g4u553 2 дні тому

    Could you possibly talk about AI developments in the VR sector ?

  • @SK80BOY
    @SK80BOY 3 дні тому +4

    Using this channel to optimize my life with ai! anyone else?

  • @supersupersocco
    @supersupersocco 15 годин тому

    I cannot generate anything usable on LTX. Anyone else generate anything usable?

  • @XenRios
    @XenRios 3 дні тому +7

    "FINALLY!" SOME GENIUS took a couple years to figure out to CODE A.I generated onion layers CEL SHEETS CREATING ANIME MOVIE, INSTEAD OF ONE IMAGE! WERE GETTING CLOSE ENOUGH TO HAVE ANY ACCESS

  • @ckq
    @ckq День тому

    Dawg we live in the future. If only someone could create an interface that has all the best image/video editing tools.

    • @ckq
      @ckq День тому

      Basically imgops but for GenAI tools

  • @gumikebbap
    @gumikebbap 3 дні тому

    18:07 waaait gpt vanilla can only use dalle, so this is kinda big!

  • @evacuatedspace69
    @evacuatedspace69 День тому

    Can you do a video LTX, it looks pretty interesting?

  • @GoblinMediaNetworkTV
    @GoblinMediaNetworkTV 2 дні тому

    But isn't 4D to do with space-time?

  • @ronyu7819
    @ronyu7819 3 дні тому +1

    AI now does 4D videos. When will it do 4D chess ?

  • @1conscience0dimension
    @1conscience0dimension 3 дні тому

    03:14 A first image. A second one. With a first IA you add an object in the hand of a character and with this one, you make the interpolation. I can't imagine all the fake scenes possible. Never trust videos again...

  • @spacegroove2896
    @spacegroove2896 3 дні тому

    29:26 the neon signs seems to be korean

  • @FaHiM58
    @FaHiM58 3 дні тому +1

    My UA-cam notification system is not working properly, so I search ' AI Search' in every 3 days to know if any video is published or not
    ..😂❤

  • @elarcadenoah9000
    @elarcadenoah9000 2 дні тому

    u forgot omini control demo link

  • @arthurfleck42
    @arthurfleck42 2 дні тому

    Omnimatte with real videos will be the ultimate deception

  • @islandofmisfityoutubers6734
    @islandofmisfityoutubers6734 3 дні тому +1

  • @KennethCollins-t8m
    @KennethCollins-t8m 2 дні тому

    Gràcies per compartir aquesta valuosa informació! M'agradaria fer una pregunta fora del tema: La meva cartera OKX conté alguns USDT i tinc la seed phrase. (alarm fetch churn bridge exercise tape speak race clerk couch crater letter). Pots explicar-me com transferir-los a Binance?

  • @DiceDecides
    @DiceDecides 3 дні тому

    4D videos??? how much longer until interactive 4D videos that converts people in a video into NPCs

  • @Kurie3D
    @Kurie3D 2 дні тому

    30:29 😁 It got farting in water accurately .

  • @adammakagassama5645
    @adammakagassama5645 3 дні тому

    07:01 the baby frat… two Time…..

  • @fizzypizzel6477
    @fizzypizzel6477 3 дні тому

    I'm gonna use generative omnimatte to make liminal space/backrooms videos by removing people out of shots.

  • @ROBOTRIX_eu
    @ROBOTRIX_eu 3 дні тому +1

  • @hqcart1
    @hqcart1 3 дні тому

    it's sad when X model is able to beat o1 (as a fact) just by copying the news with 0% authenticity.

  • @goHSLA
    @goHSLA 3 дні тому

    QwQ is doubting itself…that sounds quite human…

  • @boroborable
    @boroborable 3 дні тому

    fyi bump map is a depth map, which makes object have fake 3d detail geometry that gives fake details.

  • @kiiikoooPT
    @kiiikoooPT 2 дні тому

    4D what is the 4th dimension? That is 3D, not 4 x,y,z

    • @theAIsearch
      @theAIsearch  2 дні тому

      time

    • @kiiikoooPT
      @kiiikoooPT 2 дні тому

      @theAIsearch games have time and they are called 3d not 4d

  • @HarryClipzFilmz
    @HarryClipzFilmz 2 дні тому

    This is why I believe the quality of everything is shit, for them to implement AI into the industry, it's a chess move, to say the least. Cut back on workers, and all.

  • @dantealexander931
    @dantealexander931 3 дні тому +1

    yes

  • @mrimc
    @mrimc 3 дні тому

    #amazing 💥 💥 💥 thnx for the info.

  • @lalropekkhawbung
    @lalropekkhawbung 3 дні тому

    Let's gooooooo 🗣️🗣️🗣️🗣️

  • @lutexor
    @lutexor 3 дні тому

    Sora seems a bit disappointing if you have watched the sora trailer.

  • @HyunsReport
    @HyunsReport 3 дні тому

    4D? So turning 2D image into 3D is now called 4D?

    • @osadniasod223
      @osadniasod223 3 дні тому

      Science and marketing are two different things, nothing new here. The question is for me more about the demos, are thry easily achievable or just target possible investor money...

    • @theAIsearch
      @theAIsearch  2 дні тому

      xyz+time

  • @GeeGnebAb
    @GeeGnebAb 3 дні тому

    insaaaneee

  • @BruceLee-kf7rl
    @BruceLee-kf7rl 2 дні тому

    Intelligence is not a countable noun

  • @GrandelZeon
    @GrandelZeon 3 дні тому

    only a matter of time tell i can make my own mmo game and anime :) using AI

  • @flareonspotify
    @flareonspotify 3 дні тому

    very cool

  • @mandyandy-e8x
    @mandyandy-e8x 2 дні тому

    IMAGINE WHAT THIS WILL DO FOR THE < ADULT XXX> INDUSTRY!

  • @wholeness
    @wholeness 3 дні тому

    Omni is something I've been looking for to finally get storytelling animations done properly.

  • @GraipVine
    @GraipVine 3 дні тому +1

    There are no flaws in the 4D output. It is just using what information it is given. Give it a 360 view of the objects and it will turn out much better.

    • @markmuller7962
      @markmuller7962 3 дні тому +2

      Dude that's the whole point of using AI, not needing tools and 360 capture and such

    • @GraipVine
      @GraipVine День тому

      @@markmuller7962 Even humans won't know what the other side of the tree looks like, until they walk around to see it. So it is limited in what it can do.

    • @markmuller7962
      @markmuller7962 8 годин тому

      @GraipVine That's literally what AI is made for lmao, you're AI illiterate, it's a waste of time

  • @Hawa_Chapal
    @Hawa_Chapal 3 дні тому

    Leaky Sora...

  • @LouisGedo
    @LouisGedo 3 дні тому

    👋 hi

  • @dantealexander931
    @dantealexander931 3 дні тому

    cool

  • @dantealexander931
    @dantealexander931 3 дні тому +1

    very

  • @SS801.
    @SS801. 3 дні тому

    sora leaked omg 😭

  • @mrburns366
    @mrburns366 День тому

    Cat4d.. they're gonna use it for porn.. lol

  • @arithene
    @arithene 3 дні тому

    QwQ

  • @ddenizc
    @ddenizc 3 дні тому

    omni control is shaaaat , it does not work as advertised , abacus ask for money , fyi

  • @roxyu3384
    @roxyu3384 3 дні тому

    4d is the new 3d 😂