Home » Sora can create impressive photorealistic videos

Sora can create impressive photorealistic videos

by admin
Sora can create impressive photorealistic videos

Not that many hours have passed since the rumors relating to the possible future arrival of an OpenAI search engine, but the time has already come to return to referring to projects linked to Sam Altman and associates. In fact, the company has just launched its first text-to-video model, represented byartificial intelligence Sora.

In this regard, as also noted by The Verge, as well as as made official directly via the official OpenAI portal on the Italian evening of February 15, 2024, the model is already arriving in the hands of the first Content Creators, in the experimental phase (also to evaluate the potential risks linked to the possible release of a model of this type for everyone). However, the public can already view some demonstration videos.

You can see some of them at the bottom of the news, as well as by possibly linking to OpenAI’s official Twitter X profile. The results certainly do not go unnoticed, also aiming to reconstruct realistic scenes using AI. All this, as already mentioned, starting from a textual input.

To be clear, for the video of the woman walking in the city, the input used, translated into Italian, is “an elegant woman walks along a Tokyo street full of warm, bright neon lights and animated signs. She is wearing a black leather jacket, a long red dress and black boots. She carries a black bag with her. She wears sunglasses and uses red lipstick. She walks with confidence and ease. The street is wet and reflective, creating a mirror effect linked to the colored lights. Many pedestrians are walking“.

See also  Amazon Music Unlimited offer: Get 3 free months now!

For the moment the maximum duration of videos is one minute, but you understand well that we are referring to a potentially significant revolution, also from a future perspective. On the other hand, it is, on a general level, “complex scenes with multiple characters, specific types of movement, and accurate subject and background details“. Furthermore, according to OpenAI, Sora is able to recognize well how to insert a subject into a scene by trying to reconstruct what its positioning would be in the physical world, as well as making the characters express emotions.

Among other things, the Sora text-to-video model can generate a video even starting from a fixed image, as well as possibly “extending” an existing video or “filling in missing frames”. Of course, OpenAI itself admits that for the moment there are still some imperfections, relating for example to how physics is reproduced in complex scenes. Furthermore, it is worth noting that there are several details, already in some demo videos, that allow us to understand that this is AI-generated content. However, it seems difficult not to call this further evolution of artificial intelligence by OpenAI impressive.

You may also like

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy