OpenAI Shocks the AI Video World - Sora Changes Everything

More from Futurepedia:

Links

Summary
OpenAI unveiled their groundbreaking AI video model Sora. It's text to video, image to video, image generation, and video generation capabilities blow every other ai video generator out of the water. Sora's text-to-video creates incredibly photorealistic fully generated ai videos in lengths up to 1 minute. This pushes the boundaries of AI content creation with their advanced diffusion transformer model. RunwayML, Pika Labs, Stable Diffusion and all the others are left behind by this release. Cinematic ai videos are here.

Chapters
0:00 Intro
1:02 Text to video
6:10 Sora research overview
6:39 Image to video
6:55 Extend videos
7:05 Video to video
7:42 Connect videos
8:27 Text to image
8:47 Emergent capabilities
9:30 Limitations
9:58 Thoughts
11:52 Futurepedia

[MattStevens]

This is gonna hit the advertising industry and every production team hard. All the 30 second video ads you've seen through the years we're created by a production team of 10 to 30 people with different roles. I can't imagine how they are feeling right now.

[jurajbobka]

Welcome to Black Mirror season 7! Available everywhere

[richardflood]

This is amazing but even more scary. I'm not going to believe anything I see on the internet anymore.

[stefanosxenoudakis]

So from now on we have to ve prepare and accept that nothing is gonna be real anymore. You either see it buy yourself or just believe.

[DisposableSupervillainHenchman]

I am both amazed and frightened. Soon we won’t be able to determine what is real but for what we see with our own eyes.

[jarekstorm]

Very few movie shots last more than 1 minute without a cut. You could make a full length film by using cuts with this, and joining them together, once it’s perfected.

[jamesmartin]

Say goodbye to marketing departments 😂

[RuckFussia]

Anyone else thinking that this just further adds to the theory that we might be living in a simulation? A simulation all ran by an extremely powerful AI?

[Joepipsquiggle]

"NOTHING IS REAL"
Strawberry Fields Forever

[NikoPancakes]

This is terrifying. Not just for jobs, but for the creativity of humankind

[jeff__w]

4:37 “If you watch the bottom lefthand corner, there’s this trippy perspective shift here…”
It’s almost like an MC Escher drawing that looks realistic until you realize it’s impossible.

[RusticReflection]

One thought. In order to tokenize the world, one would have to build a virtual representation of it, in order to keep track of everything. Validated on blockchains using oracles, to make sure reality and VR synch correctly.

[nathanrussell-raby]

Not sure yet whether I'll be able to use any of this in the finished film, but I can see this massively speeding up the production of my animatics.

[MrReedling]

Amazing! Everyone will be out of their jobs before they even realised what happened.

[levoGAMES]

That's some amazing progress right there - I didn't think we were at this point already so this comes at a surprise.
Even so, there's something about these videos - apart from the obvious glitches - somehow they don't feel quite right.
But if they were to mix AI shots with real video, I probably wouldn't be able to tell.

[MetalDeathHead]

Now imagine making text to videos in VR

[EdAnde]

No matter how good it gets, it’s all subjected to individual knowledge on deciphering between what’s real and fake. The human eye is vey extraordinary, even as we get these updated our eyes will just adjust to the next evolution. Now if you were to take all this tech back to the year 1700-1800, then humans from that era may and probably would have problems deciding what’s real, that’s if they would even question if it’s fake at all.

[globaltraveller]

’Giving right command’ will be the next skillset ☹️

[inphanta]

With this, you could basically write out one of your dreams and have an AI produce a very close approximation of it.

[RyanHellyer]

There's going to be some crazy music videos coming soon.