In 20 years you'll be able to ask an AI to generate a sequel to your favorite movie and give it specifics and it'll do it, with dialogue and everything. Thing's are going to get weird over the next couple decades.
I'm not so sure it will take that long. The tech is rapidly improving at an exponential rate as all tech does. This tech hasn't been around long & it's already at this stage. A decade or less could see it making full on movies that are almost indistinguishable from cinema.
The hollywood strike is fighting a losing battle. The war is already won. In a few short months perhaps this same time next year we will have open source workflows that can take existing reference video and apply it to text video synthesis generation. Personalized AI entertainment was always the end goal. It will be realized quite soon.
How does this come even close to replacing actors and writers? AI can only replicate what has already been made, not make new stuff. Your comment tells me you know absolutely nothing about the filmmaking craft and have zero respect for people who make a living in that field.
The rate of generative video improvement is astonishing! Just a month ago the results were awful but now they're approaching decency. I can hardly wait to see the quality a year from now.
this is why I see no point in making any actual art collection with AI, cause the moment I make something, there's an update that can make it 10X better...
@@rexa2851 Artist’s work is being used to teach AI without their consent and they’re going to lose jobs. There won’t be a point in being an artist because AI can generate a weeks worth of work in seconds.
@@absolutefoot4594that's false. all artists learn from other art. it will only allow for art to go from concept to creation faster. this same argument was said about Photoshop, before that was the argument against computers, before that was the argument about cameras, the easel, pens, paint, etc. It's tiresome, progress will continue whether people want to stand in the way or not.
It was only a few months ago that text to video was so primitive and now look at where we are! I can't wait to see what we're going to be seeing in 6 months!
A few months ago it was at best funny, at worst ugly. But now some samples are simply very good. Considering how Midjourney improved in one year, we could imagine where T2V will be in one year.
It amazes me how we create a future where we will pay companies monthly to use an AI to make art, while we break our backs working longer hours, and with fewer job opportunities. It's nice that technology helps us, but seriously, at this point we will end up being useless if we ever have a blackout.
I am all over this workflow adding to my music videos and creating animated gifs where I normally would have had still visuals! Some of what I've seen this workflow put out is unbelievably stunning! I can't wait until I can feed it 3d animation.
I shared some of my own simple renders on my channel if anyone's interested. Thanks Alan for putting this new technology on my radar! I saw it here first and had to try it out for myself. It's truly insane how much this tech has evolved from a year ago. I hope to post many more experimental videos as the technology continues to exponentially improve.
@@eviv8010 This is pretty different from a deep fake, sticking a face on a diff body, usually on real video footage. This is taking a single image and generating an entire video clip from it, including making up how it should move, what it thinks they should look like from different angles, etc.
It remains to be seen if the development takes more leaps from now on or does it slow down. At this point it's still quite noticeable it's done by ai, but how long or will it ever achieve perfection?
I feel like this could be a useful tool for a base. You convert pictures into video and you try to mimic that movement. It'll help for movies or shows that are trying to do a dream or flashback sequence. It seems so unatural and odd. It could make for some interesting dynamic scenes in movies if someone decides to make a movie with movements and possibly scenes inspired by AI. It's familiar but different especially that space suit part.
Fascinating video. I hope that in 10 years, this will give the chance to create many innovative and deep Science fiction / Fantasy / Historical movies from original scenario or making excellent adaptation from well written novel (Asimov, Dick, Simmons, Reynolds, Barjavel and... Star Wars, Star Trek) better than most of the actual studios trapped in their mediocre or hysterical model.
Its easy to forget that AI is trained on quality real work and to get quality in future we can not only remix old content but also need good time consuming craft in future. With AI more can be creative but not forget the hands that feed you.
@hombacom ultimately yes, the creative arts and AI technology are going to have to find a way to coexist, AI can never replace creators coz AI can't create on its own. It needs content to remix and emulate. There needs to be new ways of compensation for artists' work.
@@gman7497 " AI can never replace creators coz AI can't create on its own" Ahahaha.... You don't know much about it, right? Your ignorance is your shield )))
What I think we will see however is an over-saturation of content with an ‘uncanny valley feel’, and a decrease in artists because the over-saturation won’t allow for reasonable pay
In the future: text-to-movie. Think of any movie you want to see and the Ai will create you your own customized movie complete with sound graphics and everything.
@@Rai2M Speech is improving pretty fast as well. ElevenLabs has a version 2 that hasn't been released yet. Even the current version guesses emotions from what it is reading, so I'd assume the new one will be even better.
Oh man the hardest thing about 3d blender was the compute power needed to render a 1 minute high quality video; with this technology a complete video can be done just by using very few angle image from my art work...blender 3d am back cause this was what may me shy away from such hobby
Im trying to get Gen2 to do anything similar to these demos starting from a MJ image, and failing completely. It wont even make anything recognizably similar to the MJ image. What is the secret??
@@bigmind2004 incorrect as soon as you add any prompting the video loses almost all data from the reference image, these are created with a reference image and no camera or animation data. It is still a nice tech though
@@bigmind2004 any one of those images could have just randomly zoomed in with no movement with the current gen-2, I have no doubt the technology will get there though.
i feel dumber having watched this. i would rather people color in a coloring book and call it theirs than to use ai i think they accomplished something
I honestly don't see the point of one AI understanding the 3D geometry of some scene to extreme details in order to generate it, then another AI trying to break down an input image with obviously less accuracy. Wouldn't it make more sense, be more efficient, and produce higher quality and much longer results, for that original MJ AI to simply slightly alter the 3D vectors of the objects it already knows are in the scene?
@@EdenHamilton just use the reference image only, have the subject doing something it will know how to animate like this th-cam.com/video/r_lwHskgsds/w-d-xo.html
fascinating what it can dream up :) how big is runway? how many params. presumably it will have to be bigger than midjourney. wild guess would be 40% bigger for comparable image quality. the training data might need to be several times bigger though
It is still missing some sort of 3d knowledge of where are the boundaries of the objects, i think it needs more work on img to 3d then make 2d animations
is this sentient, self-developing "ai" you speak of in the room right now? because human beings are the ones developing and applying this technology in case you forgot
I think you will be a literal God as a consciousness in an AI assisted world of your own in the distant future. YOU CAN DO ANYTHING YOU WANT but you will be alone - so treasure every consciousness you have right there, family - friends, etc. And always stick to your purpose and dreams..
all the hyped people here don't think about capital replacing art and jobs. 'the movie industry will change for the best, now everybody can create high-quality art'? hahaha. these people do not understand how that 'art' is born and purchased in the economy.
In 20 years you'll be able to ask an AI to generate a sequel to your favorite movie and give it specifics and it'll do it, with dialogue and everything. Thing's are going to get weird over the next couple decades.
I'm not so sure it will take that long. The tech is rapidly improving at an exponential rate as all tech does. This tech hasn't been around long & it's already at this stage. A decade or less could see it making full on movies that are almost indistinguishable from cinema.
In 30 years time, you'll be able to wake up one morning and find out that the past 2months that you experienced was just a simulation
Same for video games. Fanfiction will grow rapidly.
anime porn with myself in it too
20? Feels like 5 at this rate.
Really puts the Hollywood strikes in context
Now everybody is a director
Aren't the writers striking? Has no relation to this
The hollywood strike is fighting a losing battle. The war is already won. In a few short months perhaps this same time next year we will have open source workflows that can take existing reference video and apply it to text video synthesis generation. Personalized AI entertainment was always the end goal. It will be realized quite soon.
@@The4valanche Actors on strike too.
How does this come even close to replacing actors and writers? AI can only replicate what has already been made, not make new stuff. Your comment tells me you know absolutely nothing about the filmmaking craft and have zero respect for people who make a living in that field.
Absolutely wild!
The rate of generative video improvement is astonishing! Just a month ago the results were awful but now they're approaching decency. I can hardly wait to see the quality a year from now.
Welcome to the ML super power :-)
wooow what a time to be alive!
@@itskittyme Yes, Two Minute Papers are great!
We can Made movie with prompt
this is why I see no point in making any actual art collection with AI, cause the moment I make something, there's an update that can make it 10X better...
In the future, series will have unlimited episodes, each custom and tailored to your algorithm.
Video is getting there. Very impressive.
Impressive but equally terrifying. Imagine this ten years from now.
I don't understand why everyone is calling this as "terrifying". This is a big step at technology
@@rexa2851 Artist’s work is being used to teach AI without their consent and they’re going to lose jobs. There won’t be a point in being an artist because AI can generate a weeks worth of work in seconds.
@@absolutefoot4594adapt or die, nobody can stop this, it’s pretty simple
@@absolutefoot4594 Well it's the best time search for a new job now then
@@absolutefoot4594that's false. all artists learn from other art. it will only allow for art to go from concept to creation faster. this same argument was said about Photoshop, before that was the argument against computers, before that was the argument about cameras, the easel, pens, paint, etc. It's tiresome, progress will continue whether people want to stand in the way or not.
It was only a few months ago that text to video was so primitive and now look at where we are! I can't wait to see what we're going to be seeing in 6 months!
A few months ago it was at best funny, at worst ugly. But now some samples are simply very good. Considering how Midjourney improved in one year, we could imagine where T2V will be in one year.
This is it. The beginning of the content singularity
It amazes me how we create a future where we will pay companies monthly to use an AI to make art, while we break our backs working longer hours, and with fewer job opportunities.
It's nice that technology helps us, but seriously, at this point we will end up being useless if we ever have a blackout.
I am all over this workflow adding to my music videos and creating animated gifs where I normally would have had still visuals! Some of what I've seen this workflow put out is unbelievably stunning! I can't wait until I can feed it 3d animation.
Tbh its scary how the technology is evolving
This is amazing! The man standing on the beach is stunning btw
I shared some of my own simple renders on my channel if anyone's interested. Thanks Alan for putting this new technology on my radar! I saw it here first and had to try it out for myself. It's truly insane how much this tech has evolved from a year ago. I hope to post many more experimental videos as the technology continues to exponentially improve.
Only a matter of time before Oblivion style AI videos are a reality.
Nice collection of output. 👍
This is insane... Just in a few months, imagine how it will be in a few years
deepfake been around for quite some time, and nothing really changed there...the result is still cringy looking characters.
@@eviv8010 This is pretty different from a deep fake, sticking a face on a diff body, usually on real video footage. This is taking a single image and generating an entire video clip from it, including making up how it should move, what it thinks they should look like from different angles, etc.
It’s basically deepfake image overlapping unto video gifs gone wild version scenarios ect lol
This is amazing, I'm have to hop back on and check it out some more. It could not do this when it was first released. Awesome!
No audio? Thought something was wrong with video?
Can't wait
INCREDIBLE
GUYS WE'RE FUKED THIS IS TOO GOOD
*Dr. Alan D. Thompson... my old nemsis*
It remains to be seen if the development takes more leaps from now on or does it slow down. At this point it's still quite noticeable it's done by ai, but how long or will it ever achieve perfection?
I feel like this could be a useful tool for a base. You convert pictures into video and you try to mimic that movement. It'll help for movies or shows that are trying to do a dream or flashback sequence. It seems so unatural and odd. It could make for some interesting dynamic scenes in movies if someone decides to make a movie with movements and possibly scenes inspired by AI. It's familiar but different especially that space suit part.
Waiting patiently for personalized movies, and custom AI remakes of videogames
An dmmo rpg games ect 😅🙏🏼✌️🤪✌️
Any idea on what the correct prompts are to place into Runway Gen-2 to make it look as if they're speaking? I'm running out of ides.
Well...seems like the movie industry is about to change too 😅
For the best
just one more step, the movie will be made
Game changer this would be great for movie world building.
This is unbelievable!
Absolutely astonishing. The quality is lost a bit, but a few papers/months down the line I bet this is fixed
Yup. Let the millions of people lose their jobs, because that's "absolutely astonishing".
@@Rai2M Why would I care about other peoples jobs?
Pensei nisso há 20 anos atrás para transformar meus roteiros em filmes. Vejo que chega a hora. Tudo que penso se torna real. Tudo mesmo.
This will benefit manga animations a lot. Who else thought the same?
Wild times hurtling towards us people.
"Skynet begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th 2023."
It's slowly getting there.
Fuck yeah, so close to my Batman Beyond movie
Does anyone know the specific steps and prompt in GEN-2 to animate a still in this form ?
I'd rather wait before subscribing to Runway !
I'm ready to make movies with something like this.
In the end, the key is how to make the story interesting.
But "runway gen2" is the best tool.
00:46 isnt it interesting that the ai gives each of the girls different personalities? That is pretty cool.
no, u just feed it videos of groups of people like that and it will mimic it
It can also be stereotypical, it tends to do things that are common
What is this exactly ?
that's incredible
Esta tecnología nos va a dar de vuelta los dibujos animados con estérica 2d de una forma gloriosa y estoy super feliz de poder formar parte de ello.
IT'S TOO GOOD 😊😊😊
Who would've thought that AI would take over acting jobs first...
I get goosebumps straight away.
Fascinating video. I hope that in 10 years, this will give the chance to create many innovative and deep Science fiction / Fantasy / Historical movies from original scenario or making excellent adaptation from well written novel (Asimov, Dick, Simmons, Reynolds, Barjavel and... Star Wars, Star Trek) better than most of the actual studios trapped in their mediocre or hysterical model.
Well said about the studios.
Its easy to forget that AI is trained on quality real work and to get quality in future we can not only remix old content but also need good time consuming craft in future. With AI more can be creative but not forget the hands that feed you.
@hombacom ultimately yes, the creative arts and AI technology are going to have to find a way to coexist, AI can never replace creators coz AI can't create on its own. It needs content to remix and emulate. There needs to be new ways of compensation for artists' work.
@@gman7497 " AI can never replace creators coz AI can't create on its own"
Ahahaha.... You don't know much about it, right?
Your ignorance is your shield )))
What I think we will see however is an over-saturation of content with an ‘uncanny valley feel’, and a decrease in artists because the over-saturation won’t allow for reasonable pay
getting close for us to create our own movies AI budget
We are soooo close!!
In the future: text-to-movie. Think of any movie you want to see and the Ai will create you your own customized movie complete with sound graphics and everything.
No, because the speech is limited AF.
@@Rai2M i said in the future!!!
Forgot about that, AI speech still has a ways to go to sound authentic and nuanced.
then profits will tend to zero
@@Rai2M Speech is improving pretty fast as well. ElevenLabs has a version 2 that hasn't been released yet. Even the current version guesses emotions from what it is reading, so I'd assume the new one will be even better.
All the faces looked so scary !! Even the hair, when it moves.
i feel excite and afraid at the same time
Could be this work with animations?
Oh man the hardest thing about 3d blender was the compute power needed to render a 1 minute high quality video; with this technology a complete video can be done just by using very few angle image from my art work...blender 3d am back cause this was what may me shy away from such hobby
Now we can make our own movies 😀
Awesome....
Reality has been real silent ever since this dropped.
can you help withprompt in runway?
Very impressive
Very impressive.
Sora is the latest top AI video as of 2024.
Im trying to get Gen2 to do anything similar to these demos starting from a MJ image, and failing completely. It wont even make anything recognizably similar to the MJ image. What is the secret??
I'm looking for the same answer. I can't find where you can animate a photo without changing it...
If a common person can not replicate these videos out the box, it's safe to assume it's been edited traditionally.
Very impressive, and By The Way, weren't you the Headmaster while I was at Trinity House School?
Looks good.
Damn last week image to video was like a blur, now it's scary good 😅
Present 💝
Very impressive.
This. Is. Crazy. 🤯
were getting close to A tier movie quality
This is not as they said, this is under development, the results are not as you hope to be
The problem is with this method you lack control of the scene although it is impressive
Not if you know 3d programs and use angle images of the direction you want the video to go about..
@@bigmind2004 incorrect as soon as you add any prompting the video loses almost all data from the reference image, these are created with a reference image and no camera or animation data. It is still a nice tech though
@@bigmind2004 any one of those images could have just randomly zoomed in with no movement with the current gen-2, I have no doubt the technology will get there though.
I'm guessing there will eventually be the equivalent of ControlNet for video..
i feel dumber having watched this. i would rather people color in a coloring book and call it theirs than to use ai i think they accomplished something
AI develops very quickly. We will definitely see skynet before 2029.
Skynet not real only fiction made up movie dude gitgud ok lmao 🤣
Just wait and see, someone will make it happen. The Ai is open code, so anyone can create it as they want.
haha Amazing
Hi
I honestly don't see the point of one AI understanding the 3D geometry of some scene to extreme details in order to generate it, then another AI trying to break down an input image with obviously less accuracy. Wouldn't it make more sense, be more efficient, and produce higher quality and much longer results, for that original MJ AI to simply slightly alter the 3D vectors of the objects it already knows are in the scene?
not long know till we have a new Netflix type service that has a prompt to film style movie experience.
Actually this might still be useful for storyboarding if it can animate sketches as well
Wow, it's so good. Does this mean in 3mo it will be perfect?!
give it 6 months to be perfect and 12/18 months where you can do long form videos.
i ll give it 3 years at least to be "perfect"
Impressive
I tested it, it has not worked for me
wtf this is crazy, should i give up on 3d animation now already lol?
after 5 years or so, imagine this tech get implemented in unreal engine! rip acting career😹😹
Okay...this is crazy
I love would love to know how you do that in Gen2 because I don't get those results at all 🤣
Use a reference image with no prompt
Yeah me too lol, mine totally change the reference image and have barely any movement.
@@EdenHamilton just use the reference image only, have the subject doing something it will know how to animate like this th-cam.com/video/r_lwHskgsds/w-d-xo.html
@@EdenHamilton don't put any prompt text let the image clearly show what the animation should be doing
No wonder Hollywood is shitting their bricks.
fascinating what it can dream up :) how big is runway? how many params. presumably it will have to be bigger than midjourney. wild guess would be 40% bigger for comparable image quality. the training data might need to be several times bigger though
It is still missing some sort of 3d knowledge of where are the boundaries of the objects, i think it needs more work on img to 3d then make 2d animations
Sweet
omg this is crazy
Everyone's a gangsta until Black Mirror happened
ai is doing everything in its power to take the art and enjoyment of videography/photography etc
is this sentient, self-developing "ai" you speak of in the room right now? because human beings are the ones developing and applying this technology in case you forgot
Tutorial?
I think you will be a literal God as a consciousness in an AI assisted world of your own in the distant future. YOU CAN DO ANYTHING YOU WANT but you will be alone - so treasure every consciousness you have right there, family - friends, etc. And always stick to your purpose and dreams..
The future is here
all the hyped people here don't think about capital replacing art and jobs. 'the movie industry will change for the best, now everybody can create high-quality art'? hahaha. these people do not understand how that 'art' is born and purchased in the economy.
These are kind of... spooky.
why no tutorial?