Generative AI

Open AI releases "Sora": Creates hyper-realistic AI video from text

Open AI has now released its generative AI model that can generate realistic video through text – and can create complex movements, sequences, and people that look real.

Open AI sparks conversation with its new generative AI engine "Sora", which effortlessly creates realistic video sequences that deceive most people. When Open AI released Chat GPT, the world was amazed at the quality of results that could be obtained from the AI – now the world is once again astonished at how real the sequences from Sora are.

Sora can create videos in full-HD resolution and in various sizes and variations, with the ability to create longer clips. Sora also builds on both DALL-E and GPT models, which means that the text instructions given to the AI are better interpreted – the model understands what the user wants to achieve and executes it. Through so-called "promptovisuals" where visual material can be created by entering text into a prompt, the clip is thus created.

In the clip above – which is entirely AI-generated with Sora – we see a woman walking on the streets of Tokyo. To create the sequence, the following prompt was entered:

AI-generated.

Sora is currently available to selected creators to both test what the model is capable of, and to see if there are any issues to consider before it is released to the public. We have also previously written about the US AI Safety Institute Consortium, AISIC, a group of over 200 companies including Adobe, Apple, Microsoft, OpenAI, Nvidia, and Meta, which will now discuss the development of AI.