An absolutely insane revolution!

Hello,

This week, I got some exciting news to share with you! OpenAI has just released an AI video generator called Sora. Before I will cover Sora’s abilities, I’m going to tell you about the history of AI generated video’s first so that you can see how ridiculous Sora actually is. 

Let's start from the beginning: The earliest AI generated video I could find is a music video from a Russian singer called Jah Khalib, and it was released on October 28th 2022. This is the link to the video if you want to watch it: https://www.youtube.com/watch?v=9fSGKezAG00&t=12s

As you can see, it’s not really a proper video. It’s more like a transition between AI generated pictures. An article I found on this video said “This is a significant achievement in the field of AI video creation. Jah Khalib is a leading figure in the AI video creation community right now.”. So this was an impressive feat for that time. 

The next video is very funny, but it’s the first AI generated video I could find that actually had some smooth animations. Here’s the link to the video: https://www.youtube.com/watch?v=XQr4Xklqzw8

As you can see, this video is a little bit disturbing. But if you compare it to the previous video, you can see that the video is a lot more fluid. Everything apart from the smoothness is, as you can see, pretty garbage. This video was published in April 2023 and the AI model that was used to create this, is called Stable Diffusion. Stable Diffusion was, at its time, one of the most advanced video generation models out there and this was labeled as a very impressive achievement. But a lot of people were joking at that time, seeing how garbage this was, that AI still has a very long way to go. 

That all changed, when 10 months later, Sora by OpenAI was released. Sora blows every other AI video generator before that out of the water by a huge margin. Here is some video footage created by Sora and yes, this is not real and all the footage in this video was created by Sora: https://www.youtube.com/watch?v=MfL-Kalpozc

If you compare that with the previous Ai generated videos, you can see just how ridiculous that is. 

Now I will talk about Soras abilities. Sora has two ways of giving it an input. The first way is by a regular prompt, and the second way is by giving it an image. Sora will then bring that image to life and add motion to the image. In the video, a good example of this is by looking at the scene with the dog. On the left side you can see the image that was given as the input and on the right side you can see Sora’s output. With this input, sora can then generate a video that is up to a minute long. And there are almost no limitations. For example: In the video, you can see a scene that replicates the video game minecraft in astonishing detail. That means that the AI isn’t fixed on creating realistic scenes, but it can adapt to a lot of different rules that you provide in the prompt/image. 

So what is the takeaway of this whole article? I intentionally gave you the history of AI generated videos, so that you can see how far we have come in such a short time. I mean, if you look at the first video I’ve shown you and then compare it to the clips that Sora created, it's a night and day difference. But the time both clips got released, is only 16 months apart! This much progress in a bit more than a year is just absolutely insane, and it shows you how fast AI is evolving and that it will continue to grow exponentially. This is just the tip of the iceberg, and I'm excited to see what's to come. 

See you next Friday!