Sora: A New Smart Video Maker by OpenAI

You may be familiar with Chat GPT by OpenAI, and now they have created a new tech called Sora – a new way to make videos from text or static images. Imagine you take a photo of your product, artwork or setting and you generate a video around it? I think that would be amazing!

What Sora Can Do?

Sora is like a wizard that turns words into videos. You tell it what you want, and it makes a video up to a minute long that looks really good and does what you asked for.

Sora is great at creating scenes with lots of things happening – like characters moving around, doing specific actions, and having the right details in the background. It's like magic!

Sora doesn't just understand what you mean with your text prompts, it knows how to make characters in the video show feelings. It can even generate multiple shots in one video and keep everything looking consistent. 

Not only can the model create a video just by reading text instructions, but it can also work its magic on a still image. It can turn that image into a lively video by making the contents move accurately and paying attention to the details. If you already have a video, the model can make it longer or fill in any gaps by adding the missing frames. It's like giving life to both pictures and videos

Things to Improve

Sora is smart, but it has some areas where it can get confused. It might not always get the physics right in a scene, or it might mix up left and right. It's like learning to ride a bike – it's getting better, but there are still wobbles.

Keeping It Safe

OpenAI is talking to people from all over the world – like policymakers, teachers, and artists – to understand what they think and find good ways to use this new tech. Before Sora is released for public use, OpenAI is working with experts to test it for any problems and building tools to make sure people can spot if a video was made by Sora.

They have teamed up with experts in areas such as misinformation, hateful content, and bias to rigorously test the model and help the model reject text input prompts that are in violation of their usage policies like those that request extreme violence, sexual content, hateful imagery, celebrity likeness, or the intellectual property of others.

A release date for Sora hasn't been published yet, but I will report on it once it is available - I can't wait to try it!