information ChatGPT is moving into creating Video from Text.

[NickB]

ChatGPT is moving into creating Video from Text......and it looks Epic

"We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction.

Introducing Sora, our text-to-video model. Sora can generate videos up to a minute long while maintaining visual quality and adherence to the user’s prompt."

https://openai.com/sora

 

[Constantin S]

As someone whose been watching AI almost everyday for the past year, this is insane and funnily enough it's not even the craziest news of TODAY. Google just upgraded Gemini, and it's just unbelievable how much they've improved their AI game in just a single year.

 

[Kingslayer]

This looks amazing!

First thing that came to mind watching this is the potential for anyone in their own homes to let their imagination run wild, create their own blockbuster type movies/with crazy CGI effects for nothing, that would once cost millions.

 

[dnbknowledge]

As someone who uses AI solutions obsessively each day, color me skeptical. Especially when I think about the numerous nervous breakdowns I had working with DALL·E.

Now I'm not saying it cannot be done. Or that they are lying when stating that the AI videos on their Sora page were generated directly, without modifications.

I'm more so stating that they are almost certainly omitting to mention certain things.

For example, it would have been truly impressive if they simply had a list of 10 prompts and that's it. Those are the prompts, those are the outputted videos, that's what we're putting on the site. But something tells me they most likely had a gazillion prompts and cherry-picked the very best outputs.

The same way, I can generate truly gorgeous art via DALL·E and post it on my socials as if it's something that took me a couple of seconds to achieve... when in reality, that one amazing piece of art probably came after 123,456,789 failed attempts and three laptops thrown out the window.

 

[Alop]

Now I'm not saying it cannot be done

I too think ChatGpt will fall behind. Sora?

 

[dnbknowledge]

I too think ChatGpt will fall behind. Sora?

I seriously doubt they will push Sora in a way that competes with ChatGPT. More likely than not, Sora will be included in the ChatGPT experience... remains to be seen what (if any) additional costs there will be and how they will go about said implementation.

What I was trying to say is that while I believe tech that enables us to create videos in a smooth, quick and precise manner is coming, it's not just around the corner. Contrary to what I see a lot of people saying after watching the Sora videos.

 

[Alop]

.

 

[coyoteblack]

Kind of crazy how quick sora based domains got picked up.

 

[jpariscoder]

ChatGPT is moving into creating Video from Text......and it looks Epic

"We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction.

Introducing Sora, our text-to-video model. Sora can generate videos up to a minute long while maintaining visual quality and adherence to the user’s prompt."

https://openai.com/sora

So I can finally get real star wars movies, just have to have AI write the script haha

 

[NickB]

As someone who uses AI solutions obsessively each day, color me skeptical. Especially when I think about the numerous nervous breakdowns I had working with DALL·E.

Now I'm not saying it cannot be done. Or that they are lying when stating that the AI videos on their Sora page were generated directly, without modifications.

I'm more so stating that they are almost certainly omitting to mention certain things.

For example, it would have been truly impressive if they simply had a list of 10 prompts and that's it. Those are the prompts, those are the outputted videos, that's what we're putting on the site. But something tells me they most likely had a gazillion prompts and cherry-picked the very best outputs.

The same way, I can generate truly gorgeous art via DALL·E and post it on my socials as if it's something that took me a couple of seconds to achieve... when in reality, that one amazing piece of art probably came after 123,456,789 failed attempts and three laptops thrown out the window.

There is a spreadsheet that contains around 1 million Mid Journey prompts, I reckon in time there will be similar for this - which speeds up the learning speed quite a bit

Lot of very smart people out there figuring this stuff out and sharing how they got the results