OpenAI has introduced a new model in ChatGPT 4 called Sora, which allows you to create realistic videos. This is the feature that everyone has been waiting for, and so far, no one has introduced a video from a prompt like this, because it is in ChatGPT 4 and it is paid.
OpenAI shared short clips to show what it can do, like woolly mammoths walking through snow, waves crashing on a cliff’s shore, and people doing regular stuff like reading or walking in the city.
CEO Sam Altman introduced a text-to-video model in a post on X (formerly known as Twitter). He mentioned that this model can make videos up to 60 seconds long with detailed scenes, fancy camera moves, and many characters showing lively emotions.
It’s unclear what emotions the first video Altman shared should show. It’s a video of a couple walking down a snowy street in Tokyo, but you can only see their backs. Still, the video looks very real and matches the description it was given very well.
Introducing Sora, our text-to-video model.
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. https://t.co/7j2JN27M3W
Prompt: “Beautiful, snowy… pic.twitter.com/ruTEWn87vf
— OpenAI (@OpenAI) February 15, 2024Previous tries at making videos with AI have had mixed results. Last month, Google showed off videos made by “Lumiere,” a text-to-video model that’s better than before but still has some limits.
Prompt: “A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. she wears a black leather jacket, a long red dress, and black boots, and carries a black purse. she wears sunglasses and red lipstick. she walks confidently and casually.… pic.twitter.com/cjIdgYFaWq
— OpenAI (@OpenAI) February 15, 2024But what we’ve seen of Sora is way better than Lumiere.
In Altman’s discussion and on OpenAI’s website, you can see videos made by Sora that show different scenes in amazing detail.
Prompt: “Animated scene features a close-up of a short fluffy monster kneeling beside a melting red candle. the art style is 3d and realistic, with a focus on lighting and texture. the mood of the painting is one of wonder and curiosity, as the monster gazes at the flame with… pic.twitter.com/aLMgJPI0y6
— OpenAI (@OpenAI) February 15, 2024They range from realistic wooly mammoths and a sci-fi movie preview to a cute animated monster and a beautifully crafted paper world of a coral reef. While it’s unclear if the videos in the CEO’s thread were edited, the ones on the website, like scenes of the California gold rush and an art gallery tour, were all made directly by Sora without any changes.
But there are still some questions. How many videos did OpenAI make, and did they only show the best ones? And how much computing power, time, and electricity did it take to make these examples?
OpenAI also admits that Sora has some weaknesses in its current state.
Since this is in the first stages and this feature of ChatGPT 4 is amazing, over time, its algorithm will become very fast, which will allow us to create our favorite videos easily. Some videos that OpenAI has shared show that with the help of Sora, we can create amazing cartoons or videos.
