How does OpenAI‘s Sora work?

SORA is the newest text to video AI made by OpenAI and easily the biggest leap in generativeAI since the advent of GPT. In this video I will show you what SORA is capable of and how it works. Given that OpenAI has not released a research paper nor the model to the general public yet, take this information with a grain of salt. Sora is an artificial intelligence generating hyperrealistic videos from text, images or other videos. In this explainer I look at the current research available and try to explain what might be in the sora paper. If you are interested in a text to video ai sora this video and sora is worth checking out.

Timestamps:
00:00 Intro
00:24 What is SORA?
01:05 Prompting SORA with still images
02:14 Extending video length with SORA
02:57 Video Transitions with SORA
03:59 Image generation using SORA
04:36 Emerging Capapbilties
06:27 Can do SORA more than video generation?
07:11 How does SORA likely work?
---

This video contains advertising content.

---

Attribution:
■ All videos but one are by OpenAI

#sora #openai #texttovideoai #stablediffusion

[erank]

Nice, thanks so much!
one comment-> I believe they are using "diffusion" models as a technique rather than actually using "stable diffusion" which is a specific model by stability ai

[bl]

Very soon it will be impossible to differentiate computer video from reality