It is now possible to create a high-quality video from a text prompt using OpenAI. This can be the future of web content and YouTube. Creating up to a minute-long video in high quality would be a boon for creating a website with an animated video background for a slider or something. This will only get better and better, how long until this can create movie scenes and a budding film creator could make their movie starring the Rock and Vinnie Jones? This could be an interesting project. This technology will be restricted, and of course, will not create sexual or racist content, but for web content developers and for creating a tutorial this would be amazing. The train video below shows that this system can cope with fast motion, the closer buildings are moving fast, but the distant ones move slowly. This and the reflection in the window does showcase the amazing abilities of the Sora system. And the Old West scene. Imagine creating a movie that is Robocop in 1885 in Detroit stopping crime. The Nigeria crowd scene is also good, but the eyes are a bit strange.
There are funny glitches with this, but this will hopefully improve with better skeletal simulation and better physics. The drone camera videos are amazing. The movement of the ocean and the surf is very good, and the panning of the camera is very smooth. I mean you would need a cluster of A100 GPUs with 64 Gigabytes of VRAM to train this model, so it is not affordable to the average consumer, but for those who can afford it, this is incredible technology. However, an open-source model would struggle as it would be a long time until a consumer video card has enough VRAM and power to run such a model.
See this here: https://openai.com/sora.