Ok guys, this time you can put "shocked the industry" in your title.

@WildPeanuts

3 ай бұрын

exactly what i was thinking

@creativeprismstudio1003

3 ай бұрын

Hahaha

@DanielBowne

3 ай бұрын

I almost didn't click dude has cried wolf so many times.

@executivelifehacks6747

3 ай бұрын

Perhaps the entire industry

@phxlightz9150

3 ай бұрын

😅

I am a director and cinematographer. And I can tell you. It's game over. For the gaffer, grip, set dresser, make up artist, best boy, etc, etc, etc.

@dennislubbers4614

3 ай бұрын

and for me too.... gonna make films for fun now! That's gonna be nice :)

@siobhan6489

3 ай бұрын

As an on set medic, I'll still be ok😁... Oh no, wait, damn, there wont be any sets for me to sit about on!!🤣

@minimal3734

3 ай бұрын

As you are directly affected, I would be interested in your opinion. Do you have an exclusively negative view of the technology? Or do you think it enables new forms of expression for people who were previously excluded for technical and financial reasons?

@JohnSmith762A11B

3 ай бұрын

@@minimal3734 I'm a filmmaker and so is my daughter (she is running up hundreds of thousands of dollars in debt at a legendary film school right now). The democratization of film production has been going on for years, and this is another step, but the studios control theatrical, streaming, and broadcast TV distribution, and have the money and connections to promote their films and get their stars interviewed on Kimmel. So even if you could crank out great films all day, every day with Sora 3.0, no one would ever see them and they wouldn't even get into film festivals (the important ones Hollywood has more or less monopolized too). It's like everyone with a Mac has a recording studio now, but that doesn't mean just anybody gets to be Taylor Swift. Gatekeepers are a reality in all forms of media. Still, being a bedroom artist making movies is great fun. Just don't expect to be more than a hobbyist showing your films to friends and family and anyone on social media who can be bothered to care. Also, OpenAI will censor the living daylights out of this tech so pray the open source world builds something to compete with it where actual artists can be artists and not creatively emasculated hacks making stock footage and movies for audiences of emotional juveniles.

@cultserpent4501

3 ай бұрын

@@minimal3734 I bet that "technical and financial reasons" were not the problem that you were "excluded" from the art industry my friend, but the perception that typing prompts = art expression is the problem that you were "excluded". Trying to avoid the obstacles, trying to avoid the time NEEDED, trying to avoid the actual CONNECTION to the creation process is the reason that you were "excluded". By you i mean those "people who were excluded for technical and financial reasons". Funny and sad at the same time that you seem to believe that people should not be excluded for technical reasons.

Nothing is morphing, blending or deforming? This is the end of stock video, in an instant, and paves the way for completely personalised entertainment. This is a pivotal moment in history

@vectoralphaAI

3 ай бұрын

Imagine the progress by the end of the year or next year. AI R&D is exponential.

@kairi4640

3 ай бұрын

I wouldn't say nothing, there's still some slight inconsistencies with hands etc if you look really close. But definitely much better than last year lol. 😂 People have been also pointing out legs moving the wrong way and extra legs on the cat grow. But it's cool you can miss it now.

@IceMetalPunk

3 ай бұрын

There is a little bit. But it's far and away the best temporal coherency I've seen from any text-to-video model to date.

@360.Tapestry

3 ай бұрын

more like the unraveling of history. soon we can generate footage for the version of history we prefer

@IceMetalPunk

3 ай бұрын

@@360.Tapestry That's been possible since at least WWII. This is just another tool to do it. "But it makes it easier!" Sure, but the people with something to gain from it were already putting in the effort and resources; the people that need it to be easier gain nothing. (Except, I guess, in the case of deepfake revenge porn, but that's a whole other issue than historical propaganda.)

A year from now we're at: "Generate this movie script into a 1h long movie"

@VaneNickOke

3 ай бұрын

I’m tending to believe

@thetimecapsuleofpatrickand4582

3 ай бұрын

its the dream!

@vaporizejello

3 ай бұрын

Dudesy talked about this like 2 years ago actually. (Ep 22?) My mind is blown.

@adamfilipowicz9260

3 ай бұрын

wont be long

@TheMonkeyShowOnline

3 ай бұрын

Even something more like 'here's an idea, write the entire script AND generate it into a full length movie'. I would honestly lose SO many hours of my life giving crazy prompts and then watching the insanity the AI comes up with!

I was considered a lunatic optimistic among my friends, and I thought it would take 3 years to make this jump in quality, but it took only 1

@willdegra317

3 ай бұрын

Humans are bad at estimating exponential change. I have a bet that we’ll see a full movie totally AI generated by august 2026 (over under). Others thought it would take closer to a generation

@vectoralphaAI

3 ай бұрын

AI progress is exponential. It always improves faster than any other for of science and research.

@360.Tapestry

3 ай бұрын

at the beginning of last year, i predicted it would get this good this fast - secretly, i thought i was just being hyperbolic to be dramatic, tho

@MrCoffis

3 ай бұрын

Feedback feeds back into the previous feedback…and so on…no one can predict how this will go on. It’s going to be an explosion for sure.

@jamak919nc

3 ай бұрын

This is exactly what I’m thinking too, just imagine what the government has access to, this shxx is really about to change the world in the next five years, maybe even sooner like 2026

We’re going to be able to Prompt movies very soon! 🤯 🤯

@simonsanchezkumrich8489

3 ай бұрын

I thought maybe it was gonna be 10 years but seeing this now maybe in a few years we'll have it

@ksem1337

3 ай бұрын

@@simonsanchezkumrich8489 By the end of 2025 is will be like that. Basicly the first company to unify few complex models into cinema generator will win the jackpot, the race is already on

@CreepToeJoe

3 ай бұрын

How about whole books scanned via OCR and used as prompts to generate movies that no one ever thought of making? 💯😯

@conhuir

3 ай бұрын

@@CreepToeJoe 🤯

@executivelifehacks6747

3 ай бұрын

I feel like cyberpunk was, in many ways, less far fetched than what we are living through now.

I can easily see in the very near furture (years?), the ability to take your favorite novel, dump it in as a prompt and have your very own custom "Movie". One way to re-experience some of those books you grew up on that would never get a movie contract.

@madebymidnite

3 ай бұрын

As an Author, I can't wait!

@gabrielandy9272

3 ай бұрын

and its likely to get the full book instead o the movies "lets skip this cause its too long for a theather and we need more money"

@sirplease182

3 ай бұрын

In my version of Star Wars, Han shoots first!

@joebroart

3 ай бұрын

Imagine an ai generated Blood meridian movie 😂

7:12 “that cat is real, it is not an ai generated cat” Meanwhile the cats arm is on the bed and a fifth arm comes and replaces it.

@dannyroessler

3 ай бұрын

that realistic five legged cat... fix it in the post bro ^^

@Draconatus24

3 ай бұрын

@@dannyroessler Make the tail a sixth leg and I’m all in.

Imagine when computational power is enough to run these models live. Set up some rules and controls and you just revolutionised gaming by creating an endless interaction and possibilities video game.

@moodmaker2796

3 ай бұрын

Also having an Engine being trained by a visual/ video AI on physics in regards to different materials without anyone having to hand craft them... The video AI can just create different scenarios and the engine translates them into simulated behavior based on the already existing vector data!!

@joaquinb333

3 ай бұрын

you might as well create a reality like our own by that time

@PiyushSingh-qs8yc

3 ай бұрын

Yes we can actually make game engines on live A.I. frame generation for eg. There are live image generators like sdxl turbo we are near that.

@TeddyLeppard

3 ай бұрын

Yep. Generative gaming, customized news delivered by whomever you want, movies or TV shows which are tailored to your preferences... people get ready.

@UFOgamers

3 ай бұрын

That's exactly what I am waiting for to finish my space game.

Yes, most movies are made of many short clips, cut and pasted to a long play. With this, I can now begin production of the highly anticipated Gladiator 2, and the Real GoT ending.

@willdegra317

3 ай бұрын

Please do

What impresses me is not the high resolution or the length, both can be trivially improved with more compute, which OpenAI has a lot of. The parts that impressed me the most is the consistency in the motions and how well the model is able to follow the prompts. Yesterday, I could tell if an image is AI generated by looking at reflections, especially on semi-transparent glass, well, I can now confidently admit that there is no way I can consistently rule out AI generated video.

@LouisGedo

3 ай бұрын

Yes, agreed. Also, the capacity of the AI model to retain object permanence is quite astonishingly good. After all, humans pretty much take for granted that video we see in 2D are actually a very complex mesh of 3D objects and to get such a complex environment to look right in a 2D video is way more difficult than most imagine.

@pse2020

3 ай бұрын

These clips all have AI flaws if u look closly... But in a quick clip on tiktok no one will notice and in a year or 2 maybe we will all be able to make our own movies. For once we dont have to wait for hollywood to make a none shitty sci-fi movie yay

I was not expecting this much progress in AI video generation SO SOON. This is a different world now.. The only way to know anything for sure will be to see it in person. I don't think humanity is ready for this.. We already can't agree on objective truths. I'm glad I got to enjoy the pre generative AI era🥂

@MatthewCleere

3 ай бұрын

It will simply take us back to a pre-photography era, like the early 1800s and literally all of human history. This is why letters were sealed with wax stamps for authenticity and breaking the stamp was a crime if it was official communication. People managed just fine back then, I think, but it will be a HUGE adjustment.

I worked in the TV industry when everything was completely analog. And in my life, I never thought that in my generation I would see a video of such good quality produced by artificial intelligence. I thought we might see something like this maybe in 2025. How wrong I have been all this time. I understand that the next 5 years of AI will continue to be mind-blowing. I confidently state that actors, including those in Hollywood, have reasons to be concerned.And this is only the beginning.

@dawn21stcentury

2 ай бұрын

So you're whole life you've been thinking this would come in 2025 and it's here in 2024. I'd say that's quite spot on.

So, I'm assuming stock footage just like stock images will be completely replaced...

@stacksmasher

3 ай бұрын

Yes. Also movie stars and movie studios.

@nickgirdwood3082

3 ай бұрын

​@@stacksmasherI doubt that.

@stacksmasher

3 ай бұрын

@@nickgirdwood3082 It's already happening. Only 10% of what we built is public.

@jakeskate29

3 ай бұрын

Yes, some people who would have bought stock video will just create their own but many people will still buy stock video. Creating your own videos/images takes time & skills, so many professionals would rather use their time to work on their own fields and leave the video/image generation to the artists and prompt experts.

@memegazer

3 ай бұрын

I don't think so completely. In the commercial mainstream it might be used to supplement, but unless this vid generation has high quality fine control and consistent results it is unlikely to completely replace stock footage imo. But for more independent projects this will be very disruptive for sure. It will definitely change the game for social media content creators I believe...but I think there is still "a few more papers down the line" as two minute papers says, before the mainstream pros of the industry have to worry about their jobs.

15, 30 or 60 second advertisements created using generative video are now completely possible. This is outstanding.

@propershroper

3 ай бұрын

So will fake news footage. This isn't a good thing.

@LuckyLucky-pc3tz

3 ай бұрын

​@@propershroperthere nothing we can do to stop so know how to control it eventually.

@Cameronations

3 ай бұрын

This isnt a good thing. Many people will lose their job and people could now spread misinformation easily.

@user-pv6vd3ug5k

3 ай бұрын

AI still has no sense for design, it is still random generic boring and just not good, beacuse there is no real intelligence behind the creations

@tracy419

3 ай бұрын

Considering I've read this exact comment already, you could say the same for what people do.​@@user-pv6vd3ug5k

7:14 It cracks me up you said it's not an AI generated cat but then it grows an extra paw 😂

Im convinced that in 2 to 4 years from now, game environments will be able to be typed into existence. This is wild tech. Truly another mind blowing technology is graduating from its infancy stage

@Yomi4D

3 ай бұрын

Spoken not typed. Then thought into existence

@ashdang23

3 ай бұрын

Art - Video/Movies - Games - VR We’re just polishing off art and it’s nearly done and like almost there then everyone will shift over to video which we’re already seeing. we will probably get short films by the end of this year or by the end of 2025. Games will come straight after

@360.Tapestry

3 ай бұрын

let's all hop in the holodeck... computer, resume my last simulation on the ocean floor

@Vartazian360

3 ай бұрын

@@ashdang23 theres already short films by ai on KZread. Though nothing of this quality. Most of them are creepy, dreamlike videos. Look up ai carnival of the ages

@cultserpent4501

3 ай бұрын

thats amazing, that way you will be able to dive deep into virtual reality and forget about your problems and your life faster and longer until you are a grandpa and realize you wasted your opportunity to experience real life and instead you remained hidden in your VR fantasy world..

When I saw the Wolf clip on the demo site, my immediate thought was "damn... they made it as hat tip for Matt, didn't they... and I'm sure he's gonna love it!" 👌

@aegisgfx

3 ай бұрын

Why is anyone celebrating this technology, it will put the movie studios out of business, the movie studios employ hundreds of thousands of people what are all these people going to do for jobs? We can't all have a KZread channel, we can't all be making a hundred versions of the Avengers movies everyday because there'll be no market for it.

I think we're gonna see a new era of movies. Now instead of every year getting 12 movies with maybe one or two of them being good. Every year we'll get 10,000 movies with maybe 12 of them being good. As every high school student becomes a Hollywood movie producer.

@damirregoc8111

3 ай бұрын

What planet are you living where they make only 12 movies a year? Here on Earth, they make thousands every year.

@ErikLiberty

3 ай бұрын

I wanted to make a basically no-budget amateur movie but couldn't afford to hire extras as background characters. I bet I will eventually be able to add extras into my shots during post-production.

@bmgtv1116

3 ай бұрын

@@damirregoc8111 thousands of year! Rookie numbers! I want every high school student making full length movies with only slight more effort than a tiktok! Just like how every girl with a phone became a photographer!

@willdegra317

3 ай бұрын

But if you have the ability to just generate a movie at home why would you bother browsing 10000? Do you listen to SoundCloud music?

@bmgtv1116

3 ай бұрын

@@willdegra317 listening to SoundCloud. Let's not get ridiculous. You'll probably just find the few creators you like and subscribe to their stuff. Think of it like the unholy spawn between Hollywood and KZread Where the next George Lucas is some 12-year-old kid in Idaho.

7:19 that cat just grew another arm as you said it was a real cat 🤣🤣🤣

@lepfitz627

3 ай бұрын

The interesting part is that if you're not used to watch out for details that don't make sense, you don't recognize them a lot of times. I'm a graphic and media designer and I see classic photoshop retouching fails in ads constantly when others only see them if I put my finger on it, because I am trained to spot those mistakes. I bet, most people have followed where the cat's leg halfway disappeared behind the pillow and not how the fur on the cats body already starts to slowly morph into some hidden fifth leg which then also changes places with the leg in the front just before the next fast movement. But: If you check out how far video generation has come in such a short time and compare it to the development of ai image development - this is insane. And when (depending on the tool) 5-9 out of 10 generated images get the amount of fingers on a human's hand correctly by now - continuity and morphing fails this obvious might be gone in a year or so. This really blows my mind. Still: Super funny that this comment was on spot.😂 @Matt: Love your vids. Really helps me keep track of all the crazy stuff that's going on these days.

About a year ago or so, I commented that we would be able to make full movies like this in the future, on this channel.. Even through i knew it back then, seeing this starting to happen is still insane to me.

The AI not only gets the particle physics correct (smoke, water, dirt), it even gets hair & fur to act correctly to wind and gravity when attached to a human or animal. Amazing!!

I think I speak for every single person watching this, on day one, when I say, “GOTT DAMM!” 😮

@tobiaslislevatn6212

3 ай бұрын

Mindblow, after mindblow, after mindblow

@BDLabs2

3 ай бұрын

That’s a GYATT

@mrthegrievous

3 ай бұрын

This is a HISTORIC moment! Am I dreaming or is this real? I can't even imagine the next 5-10-15 years. Advancement will accelerate at a more and more breakneck speed!

@joeteti4958

3 ай бұрын

Gott dayum bro is right

@EsseQuamViderity

3 ай бұрын

You don’t speak for everyone.

Completely MENTAL. Just saw there whole demo page, and DAMN. I though this quality would only arrive in a few years and here we are already... Incredible, TRULY.

@AIVersityHub

3 ай бұрын

Sooo gonna change our world

I checked your channel an hour ago to see if this video was up yet after a friend shared a link to Sora. Incredible tool. Say goodbye to stock video services!

@lipwood_music

3 ай бұрын

Yes and no, if you need the location to be 100% legit because of details, you still need videos, but for generic ads yeah of course all the way

@skierpage

3 ай бұрын

​@@lipwood_music Unless you're needing the Eiffel Tower or Venice, a stock video footage company is unlikely to have footage of your desired location. Also, all the tools from image generation like reference images and only generating pixels within a mask are likely to come to AI video generation.

these are crazy times we living in! we are blessed to be experiencing this. Soon anyone will be able to make a whole movie with just a text input.

I noticed a lot of clips look like they are moving backwards. I suspect Sora might start with the final frame and work its way backwards to increase consistency. And sometimes it spills into the final render.

The quality is impressive! The thing I've been most looking for is being able to describe and then it create something unlike anything that's previously existed. For example, if Lucas had never created Star Wars, would it be able to come up with something like Chewbacca if described well enough? The duck-dragon (a "drake" perhaps?) is another step in that direction, but we could see the problems: more cartoony, camera made it look as if going backward, a bit dreamy. It seems the further we prompt away from reality, the further all parts of it look from reality. Then there's the problem of having that character look consistent in multiple scenes. It's much easier to with something from real life or something imagined that's well documented. It's not just that I'd like to do this for some projects, but that is actual creativity. LLMs currently generate things that have never existed by putting together things that are like what has existed before. It's a much harder and more creative task to consistently generate something which has never before existed. And maybe that will always be a human role. 🤷‍♀

@OluwadiyaKehinde

3 ай бұрын

AI will get there. If it took one year to make this much progress, imagine how much more progress can be made in five, ten, or twenty years. And the big elephant in the room, Artificial General Intelligence, is still on the horizon.

@RubelliteFae

3 ай бұрын

@@OluwadiyaKehinde Sure, but it's a different task to what's being worked on and I for one don't know how you go about working toward a solution. I've definitely gotten Midjourney to make things close to what I had thought up, but it takes 1/2 a months credits to do so and then without any consistency. So, not hopeless, for sure.

@OluwadiyaKehinde

3 ай бұрын

@@RubelliteFae I get you. The point I am making is that it is still early days for AI, and that future iterations may likely do what you are talking about...

I can't wait to see what the Star Wars fan projects do with this stuff in another year. It's almost there.

@JarJarBricks01

3 ай бұрын

The problem is I already know they're gonna blacklist certain words like "TIE fighter" "lightsaber" etc. And of course none of the actors will be allowed. So really, this will only be helpful for establishing shots.

@nikolaizaicev9297

3 ай бұрын

@@JarJarBricks01 Only till the moment when open-source version will come out with no restrictions.

@JarJarBricks01

3 ай бұрын

@nikolaizaicev9297 Correct. But those are trailing behind in terms of quality. Thus why it could be 5 years from now instead of within a year.

@willdegra317

3 ай бұрын

No one will care about actors because “acting” won’t be a thing.

@Ghostman2122

3 ай бұрын

@@JarJarBricks01 Also they would never allow anything even vaguely violent to be generated (I can understand why but it still sucks)

Here we go! I'm curious how upscaled these videos are, if at all. I can' wait!!

Next step (except getting access) will be to get character coherence, or whatever it is called to keep getting the same "actor" consistently. This already is insane though, can't wait to try it out

@marksutherland774

3 ай бұрын

yep this will be key, the option to "save" a created character, it'll come.

1:37 right leg becomes left leg 😅 . It does look promising and before you know it they improved it even more.

@eannane8712

3 ай бұрын

At least you know it's real ai that way. It looked so real I wasn't sure other than her walking being a little weird.

@Diabloto96

3 ай бұрын

wow that one was easy to miss, I noticed a few other ones though (third cat front paw, people walking into railways and disappearing, and a lift car rolling through objects) it does make me feel like this is all patchwork understanding of "real physicality" and that it lacks something fundamental. The rules it is following are almost inscrutable but are only an approximation of reality. Not a problem for fiction video, but for things being touted as realistic, it makes me a bit uneasy.

@timber8403

3 ай бұрын

The giveaway also about this clip is the fact that all the people on the street are walking at the same slow pace.

Its not even 2025 yet, and were getting stuff this advanced! Truly mind blowing!

@Gerlaffy

3 ай бұрын

Singularity

imagine fandom, making their own movies, stories, are even alternate endings to movies and TV series, I can just see people making their own last season of GoT, or, making Star Wars, stories, I also know there is going to be a lot of legal issues to be cleared up down the road. when a fan created character looks identical in quality to a Hollywood, make character.

Oh man! incredible! Wish I had time to watch all your videos, but glad I watched this one. I've shared your channel with a bunch of people and some are already subscribed.

🎯 Key Takeaways for quick navigation: 00:00 🚀 *Introduction to Sora's Leap in AI Video Generation* - Discussion about the significant advancement in AI video generation with OpenAI's introduction of Sora. - Highlights of Sora's capabilities, including creating 60-second videos with detailed scenes and complex motions. 02:06 📽️ *Showcase of Sora's Video Generation Examples* - Presentation of various examples from OpenAI's website demonstrating Sora's video generation capabilities. - Detailed description of different scenes created by Sora, emphasizing its realism and detail. 05:08 🎨 *Exploring the Advanced Realism and Versatility of Sora* - Discussion of Sora's ability to understand and translate complex prompts into realistic videos. - Examples illustrating Sora's versatility in generating diverse and detailed scenes. 07:28 📈 *Potential Impact and Future of AI Video Generation* - Speculation on the potential impact of Sora and the future of AI video generation. - Comparison of current capabilities with past AI video generation tools, highlighting the rapid progress. - Discussion on how AI tools like Sora could empower human creators rather than replace them. Made with HARPA AI

Wait, I have to check if this is April 1st.....nope. I still don't believe it, it's just too incredible. Can't wait to see you give it a try.

@1:37 you see her legs rotate around.. Nearly broke my mind!! 🤣

@skierpage

3 ай бұрын

She's a really good dancer. After years of training she evolved Michael Jackson's moonwalk into the orbit walk.

@pchungvt

3 ай бұрын

this actually proved the video isn't fake. OMG

@kairi4640

3 ай бұрын

Nice catch, I could tell something was off but missed this lol.

@ZoOnTheYT

3 ай бұрын

Nice catch!

@osamabinballin5369

3 ай бұрын

Still insane, nonetheless

Can't wait to be able to prompt movies, I have tried to write a complex first person present cyberpunk without any narration and I got stuck when trying to write long action scenes. I intend to make it into a VR movie/tv series maybe once AI is good enough to prodduce my vision. Can't wait.

@willdegra317

3 ай бұрын

When it’s that good few people will wanna see your series. You’ll be able to upload to a SoundCloud like site for AI videos though

If AI generates the script for a movie and Sora generates the video from that script then it all comes down to the user managing to have a good idea. Welcome to the future of television/movie.

@willdegra317

3 ай бұрын

And if everyone has an AI generating TV why bother waiting to see what some people produce hoping it’s something you like? Do it yourself. If you’re not confident in your ideas download “model based on director xyz’s films”

@rfxtuber

3 ай бұрын

but is it really generating the images and not just taking from stock video that already exists and doing kung fu on it in real time?

@Gerlaffy

3 ай бұрын

You clearly don't understand how any of this works @@rfxtuber... No, this isn't just "taking from stock video", it's literally making this based on a prompt.

@lamsmiley1944

3 ай бұрын

I envision the process being going to chatGPT and asking for a script, from there it goes to Dalle to create storyboards (that you can edit). After that it goes to Sora to generate the film based on the script and storyboards. The final step would be audio generation. It’s technically possible with the tech available today, but we probably won’t have access for a while.

@rfxtuber

3 ай бұрын

@@Gerlaffy Well.. i'm not convinced on that...sure its generating video from txt prompts i understand that, but how its generating the video from txt prompts is not from thin air, it looks like to me its using pre-existing video stock footage to generate its video.. which is a very different proposition to me? Still dam awesome and unreal... but i don't think its generating the video purely from AI....Its referencing pre-existing video then doing kung fu on it... How would you prove me wrong?

indeed, absolutely mindblowing! Imagine what kinds of imaginations will be set free by individuals without any equipment besides a computer. can't wait to use Sora!

I found myself looking for flaws in the videos, and they are definitely there if you look closely. But at this point they are more of a relief for me, because it means these are actually AI video and not a prank by OpenAI. Crazy good quality.

@MemphisSweden

3 ай бұрын

The orange cat in bed lift a second left arm/paw when touching the woman, having two left arms. :P also the woman walking in the city kind of glides when she walks if you look closely. Just small quirks. Otherwise it’s mind blowing and amazing.

not knowing the first clip was generated by ai and you mentioning it was absolutely blew my mind. it's incredible how big of an improvement there's been

I havnt been so impressed by the improvements lately, until now, this is feels like a big step forward.

No one could have predicted that this would happen in 2024. Absolutely incredible.

@temet0nosce

3 ай бұрын

I think some people could of predicted it, considering that we already had ai video pika and runway and this year barley started.

@BlackTakGolD

3 ай бұрын

​@@temet0noscepika and runway here is like comparing gpt4turbo to lama2.

@thehandleiwantedwasntavailable

3 ай бұрын

Yeah, some of us have predicted it.

@user43712

3 ай бұрын

and its just February 😳

@channeling764

3 ай бұрын

People have no idea how fast this technology grows. I wouldnt be surprised at all if by the end of this year we’ll see an AI movie. Singularity is coming faster than ever.

Finally someone who properly uses the "blew everyone's mind" - in this case it really is fitting feeling. Thanks!

So sic, so pumped for this! Tik-Tok about to get wild. I wonder if we can ask for different timestamp scenes.

This really shocked the industry. 60 second Full Hd 1080p video. You only need 60 prompt to make 1 hour movie

GG, Pixar

As a visual effects content creator for 25 years, it’s bitter sweet progress. Great to democratise the creation of this stuff so more independent ideas are born. But kinda shit that it’ll decimate my industry in the process.

@DUDEiTHINKIMNUDE

3 ай бұрын

Can you start your own business and make videos for people?

@leviwakeham

3 ай бұрын

😂

@minimal3734

3 ай бұрын

I understand the feelings of those directly impacted. But I believe there are no "artists" and "ordinary people". As you say, this tech removes the irrelevant technical and financial barriers which have prevented people from expressing themselves in the past.

@JohnSmith762A11B

3 ай бұрын

@@minimal3734 Which is great, but no landlord is lowering the rent so you can stay home and make movies in your pajamas. AI is going to decimate an already devastated working class. We need UBI and we need it yesterday.

@user-be4tl8nw4r

3 ай бұрын

​@@minimal3734there is no ordinary people and artists, only people who are lazy, don't have time or money to learn and people who work hard no matter what

I actually think continuing to photograph the world will be more important in the future so the AI can easily recognize places that it may have trouble with. I want to be able to prompt a specific street or place and get a somewhat exact portrayel of what it looks like in real life.

Thank you, Matt! Good information!!! :)

People will get upset at artists that are upset about this and say “it’s just a tool”, but isn’t a tool meant to assist an artist? This literally just replaces the artist. Once this is fine tuned in 5 years all you’ll need is screen writers to give them stories… oh wait, it could probably do some screen writing as well. I seriously don’t understand why people love this so much, it will take the soul out of art.

@SamAttenborough

3 ай бұрын

Because art nowadays isn't about soul. Its about consumerism

@adamwright4135

3 ай бұрын

To a large part, the art is usually in the idea, and humans still have a monopoly on those

@redmictian

3 ай бұрын

No. I will democratize the art. Everyone will be able to create, not only those who has money or years of experience. That’s the new era and that’s has progress has been working. We will have some many talented people expressing themselves, not just a tiny few for money

@lorenzovecciosegniore1212

3 ай бұрын

When photographs came, everybody said that paintings are dead, and when movies came, everybody said that books are dead.

@minimal3734

3 ай бұрын

There are no "artists" and "ordinary people", and this technology removes the irrelevant technical barriers which have prevented ordinary people from expressing themselves in the past.

I knew the tech was advancing fast, but this is mindblowing

Thanks for posting. Wish they were releasing this next week.

This is an incredible advancement! Now in order to make this truly useful for production, we just need a version with Video2Video so you can direct your set layout and camera motion, and a style frame for art direction.

Can you link videos to make an entire scene? Or does it generate from scratch each time? Wondering how long it will be before we see full length movies using this.

@ErikLiberty

3 ай бұрын

AI pictures are now able to have the same consistent theme if you want rather than generating a new one from scratch. I bet the same will happen for AI video eventually.

We're going to be lucid dreaming in our VR goggles while we're awake

@MatthewCleere

3 ай бұрын

Exactly. Combine this with tech that comes from studying Halo, and the new "screen" will be inside our heads eventually. Jacked into the matrix. The implications are staggering and potentially dangerous to use.

@cultserpent4501

3 ай бұрын

while you will be dreaming in your favorite VR goggles, your masters will further control your REAL life without you even noticing. Until everyone ends up slave to robot masters.

Absolutely mindblowing! I was left speechless for half of it. This is going to change creativity in so many ways going forwards. It has only taken one year to get to this level... unreal 😲

The reflection-in-the-window sampling is super complex. Incredible to render an effect like that. SO(a)RA is truly going to leave no eyes sore. 😁

7:15 two left paws

@blaspayri

3 ай бұрын

7:05 the second left paw appears

@xviii5780

3 ай бұрын

Man that's a very high level problem to have right after "cursed Will Smith eating rocks"

@channeling764

3 ай бұрын

I still can’t even notice. I don’t think people would care about details. Companies just want results that are quick. They don’t care if it comes out perfect. If it takes time, it’s out.

Sam: Hit me with your best shot. Me: Cinematic long steady cam shot, following a well dressed couple in 70’s New York mafia dress, entering the service entrance of a busy Italian restaurant, being led through a busy kitchen, past a big band on stage and taking a seat at a reserved table surrounded by other mobsters, in the style of Martin Scorsese.

@JohnSmith762A11B

3 ай бұрын

Prompt rejected for the words "Martin Scorsese". Read the Sora site about IP.

Wow! I’ve been making long take shorts using Runway Gen-2, and as an experimental filmmaker am embracing the AI morph effect. I think there will be different applications for different styles of filmmaking moving forward, which is super exciting for filmmaking, and will provide the opportunity to expand on our genres!

This particular section "In this work, we find that diffusion transformers scale effectively as video models as well. Below, we show a comparison of video samples with fixed seeds and inputs as training progresses. Sample quality improves markedly as training compute increases." articulates how far video gen has come!

This is exciting! 🔥

@brobunni

3 ай бұрын

Exciting and scary!

WHOA! But the cat had two left paws! That's so weird!

Great video Matt. Take a look at 8.35. The grandmother birthday party.... Would you agree that the younger woman in that scene to the left of grandma actually looks like she is related? It'd be interesting to see the prompt for this to see if the author put that family resemblance in the prompt as an instruction or did the AI just do it as a matter of course? because, if it did I think there must be some AI driven reasoning and thought behind it.

WOW SUPER DOPE! I think the best part is how excited Matt is. Rock on Big Dog!

Im completely convinced OpenAI has access to AGI, its the only way they can completely jump/skip 10 years ahead of everyone else in video generation. Pika and Runway are 10 years behind this tech. This is total insanity, my mind is completely blown away. This should not be possible at all yet in 2024, this is a 10 year leap in AI video generation; unless OpenAI is using the power of an AGI to increase productivity 1000x fold.

@weightelk

3 ай бұрын

Sheeeeesh it's insane

@WaveOfDestiny

3 ай бұрын

The way the prompts are executed so realistically makes me believe we're much closer to AGI... Damn even if i try to immagine these scenarios in my head, i would totally miss some details

@Bilangumus

3 ай бұрын

They have billions of dollars and massive CPU/GPU centers.

@skierpage

3 ай бұрын

Don't exaggerate, this is a 9-month advance because things are happening so fast. And maybe OpenAI just threw 50x more compute at the problem.

@dannyquiroz5777

3 ай бұрын

🤫

"Give me an episode of Frasier but with the cast of Friends with a little bit of Family Guy humour..."

@vxery

3 ай бұрын

Brilliant

@cleano2702

3 ай бұрын

I'd watch that.

@RelaxingSounds-sb3hq

3 ай бұрын

An episode of FRIENDS but with all-black cast

Thanks for the update! So cool

I'm speechless! Thanks for keeping us updated. I have one question, what will happen to all the Stock Footage Websites?

As a software engineer and a heavy Midjourney user there have not been many times in my life when I could not come up with the words to describe what I was seeing but WOW, the only three words I kept repeating were, WTF! As most have been saying, what will the next 3-5 years bring? And as the guy from two minute papers keeps saying, "What a time to be alive!"

@JohnSmith762A11B

3 ай бұрын

Kinda the last straw: I'm selling all my indie filmmaking gear as shooting stuff in the real world is now an expensive hassle when I can get better results with a home computer and tech like Midjourney/Automatic1111 and Sora (soon, I hope). If I wanted to replicate any of these Sora samples barring the cat video it would take me weeks or even months and cost a fortune. No wooly mammoths available at any price either...

I wonder if they did the same thing as Microsoft at the Superbowl and just rolled a million times for the best possible video on their best private model and then showed that one just like it was the very first result

@titannb9027

3 ай бұрын

its obvious. but at some point it will generate a good 1 min video.

@tobiaslislevatn6212

3 ай бұрын

They also showed bad results, probably wont get a perfect video each time, but still look like you can get some pretty good videoes

@skierpage

3 ай бұрын

SamA responded to tweeted suggestions with generated videos pretty fast. (As an aside I wonder what it's like to have the full power of an uncensored multimodal AI foundation model available 24x7. Sam probably has the faked Google Gemini video working for him, constantly digesting his audiovisual feed to interpret everything he sees, hears,, and says. It's the movie Her for real.)

@leahgale

3 ай бұрын

@@skierpage this is a good point! :) Yes perhaps openAI will be more honest about how they handle things.

@leahgale

3 ай бұрын

@@tobiaslislevatn6212 yeahhhhh

This is good news for creators and recipients. Thanks to the fact that the creators will have almost no work, they will not have to earn as much, which is why I hope that KZread and the creators will reduce the amount of advertising. No more KZread Ad blocks KZread needed,crazy.

When people started talking about being in a simulation, they were talking about rendering a GTA like world with textures the size of the universe. Compute power would require some quantum level breakthrough. Imagine being able to use this tool to create 3D environments procedurally so that a car generates just before you look out the window at it. It makes living in the matrix that but closer.

06:28 was the one I fell off my chair 😳

Super exciting where we will be a year from now! What an incredible time to be alive!

I don't even know what to say! This is amazing! Now I understand why so many new data centers are being built in my state. Going to need a LOT of rendering performance to make this kind of stuff!

Wow, you’re really fast! It literally got announced just hours ago!

The moment that we could create a movie from a book or a text will be incredible.

Imagine the corn it can generate

@ackerwoman16

3 ай бұрын

Pretty sure if Open AI is doing so much behind the curtains then so are the regulatory bodies in the making

It will be interesting if you're able to save characters then basically insert them into your film. Mindblowing how far they''ve come in such a short time frame.

Great update. I can't wait to see what's created with these superpowers in video gen.

Will video lose it's value though? It just feels a little like a company announcing that they figured out how to make a million tons of gold every day! Everyone can have gold now!

I guess we not suppose to talk about it but....this will take porn to a whole new level

@justaperson9155

3 ай бұрын

Nobody should watch porn.

@minimal3734

3 ай бұрын

@@justaperson9155 Nobody should think about things like this 😄

That's pretty impressive! Hope we get access soon! Can't wait to use this.

I'm a videographer/editor (Indie filmmaker) and I'm with Matt in regards to the future of videography and filmmaking. I'm not gonna lose my job, because I'm currently - like all of you - playing around with and learning about all these new Ai tools we now have at our disposal. 🤖+🕺=😎

People are already sick of fake content, authenticity and substance will be key to earning the attention of audiences over the next 5-10 years. As a videographer, I’m not worried.

@coldlyanalytical1351

3 ай бұрын

Err ... OK.

@williamparrish2436

3 ай бұрын

Hahahahaha... you are so wrong its incredible. No offense, but we have seen what all the "real artists" have been producing. This is better than 99.99999999% of what is already out there, and its about to be in everyone's hands.

@gregxcelente4271

3 ай бұрын

@@williamparrish2436 who hurt your feelings?

@coldlyanalytical1351

3 ай бұрын

@@williamparrish2436 Most 'consumers' of art - such as advertisers - will be happy with cheap 80% OK art.

@deadlydiminuendo2161

3 ай бұрын

The end.

I am deeply impressed. This is all moving so very fast.

I am so stunned by the progress made in such a short time. This is truly awesome.

I've been waiting for this for so long. Its probably the thing I'm most excited for. We're still quite a ways a way from what I'm wanting but this is a great first step. I ain't even trying to use it for profit. I've just always wanted to be able to type out my very own episode of the Simpsons, south park, American dad etc and just be able to have ai bring that to life for my own personal viewing.

Also in the five grey wolfs video they start as three. But these videos look amazing and I'm blown away

SOOOO awesome and your point about how movies cut shots now.. to make them short scenes.. plays right into modern KZread editing as well. So I predict that the Oscars will HAVE to have AI Generated Movie Categories in the coming years.. and maybe even an entire AI Awards show all by itself.. coming soon in a Future Near You. As with any tool.. Please Use AI responsibly and we can ALL get to play.. Kaiber can do up to 8 minutes now.. so, I am sure Sora will be able to match that soon.

Wow amazing --- going to be waiting for this everyday.... Great video Mr. Wolfe.

Fantastic! Yes, as film maker since 1968 I need to know: Can characters ble "repeated", so you in the end can make a "full feature"? Can you bring in own characters?

Wow, the moment we can can make image-to-video with art from places like Midjourney and Leonardo and put it into this thing, i am freaking out!!

This will make people question the reality of all digital content. From now on, nothing is for sure real until you see it with your own eyes.

We all know the most important thing is if the model can be consistent from having the same subjects in different scenes. We can’t make movies with characters morphing into someone or something else that doesn’t represent them

you can take movies, use them as training models (similar to how you have a collection of images), and create new scenes with the characters... add in voice AI and the possibilities are endless