Sora Review: Does OpenAI’s New Video Generator Live up to the Hype?

There is no denying the fact that video is the most popular form of content on the Internet today. It wields the power to captivate your audience in the moment and communicate your message more clearly. However, creating engaging video content takes effort and expertise. Well, in the age of AI, that could not be further from the truth. 

Today, we are reviewing a tool that has been making quite the buzz recently. It’s called Sora, from OpenAI - the company that created ChatGPT. 

Sure, OpenAI has a solid track record. But let’s find out if  Sora is any good and can live up to the expectations.  

What is Sora?

Sora is a text-to-video tool that runs on an AI algorithm. Developed by OpenAI, it has the capability to turn written prompts provided by the user into AI-generated videos that can run up to a minute in length.  

This sophisticated tool has been designed for a multitude of use cases in industries like filmmaking, marketing, content creation, graphic design, and beyond. 

Photorealistic AI Video

You are probably familiar with OpenAI’s DALL-E which can generate photorealistic still images from text-based prompts. Well, now imagine using similar instructions but instead, the AI model generates moving images or videos for you! That’s what Sora does.    

Looking at some of the samples published on the official website, it is easy to see Sora’s capability to generate short-form videos with any number of characters, detailed 3D scenes and materials, photorealistic lighting, shadows, reflections, and complex camera shots. It seems to have a profound understanding of language as well as how things interact in the real world.

Being built on previous research in DALL-E and GPT models, Sora is really good at interpreting what the user has requested in the prompt and can create interesting characters with strong emotions. Also, it can generate videos where these characters feel real and consistent throughout different shots or scenes. 

But wait, there’s more! In addition to whipping up videos from lines or paragraphs of text, Sora can also breathe life into still images, animating them with impressive precision and getting those tiny details right. And if you've got a video with gaps or need to stretch it out, this AI model can extend it or fill in those missing frames like a pro.

It is effortless to spot the difference between Sora and its competition in the market. So far, all the text-to-video AI systems have had a characteristic image-morphing flaw as the video progresses in time. But Sora has almost eliminated it. In videos generated by Sora, objects remain stable even when they are obscured by other objects in the foreground.   

Accessibility

Sora is still in its very early stages, which means limited access to the end user. Thus far, the OpenAI team has made it available to a handful of visual arts, filmmakers, and graphic designers so they can collect feedback and improve this AI model for use by advanced professionals in their creative work.    

Not only is OpenAI critically examining Sora for potential risks but also sharing the research progress with people outside of the company to get constructive feedback.   

Limitations

As Sora’s capabilities have only been demonstrated by a handful of public users and some samples uploaded by the internal team on the official website, it is quite hard at this point to judge how robust this AI model is. But it is good to see OpenAI admit to some of its limitations right off the bat.

According to the company, Sora may sometimes find it difficult to accurately simulate the physics of complex 3D scenes. For example, in a scene involving a person taking a bite out of a cookie, you might notice that the cookie does not have a bite mark afterward. 

Besides, Sora’s AI model may struggle with the spatial details of the instructions provided by the user. For example, it might be confused with left and right. Also, it might find it challenging to precisely create events that progress over time, for example, any complex camera motion where the camera follows a specific object or path.  

Safety Concerns

Well, safety concerns are nothing new when AI is involved. We are all familiar with an ocean of deepfake AI-generated videos on the Internet and the negative impact they could have on the person associated. But as the company claims, they are taking necessary steps before making Sora a part of publicly available OpenAI products. 

OpenAI claims to be working with experts in domains like hate speech, misinformation, and bias. Plus, they are developing tools that can detect misleading content and identify videos created with Sora online.   

Pricing

Being an early-access tool, Sora does not have any pricing yet. However, considering the monetization of ChatGPT, it won’t be surprising to see OpenAI making advanced versions of Sora available on monthly subscription plans. 

Final Thoughts

At the end of the day, Sora is quite an impressive tool that can streamline the workflow of creative video professionals. Despite its limitations, its capabilities demonstrate how much generative AI has improved over a span of just a few years.   

Just like the DALL-E and ChatGPT models, Sora will only get better as it becomes public and collects feedback from the end user. As long as the company can find a way to combat deepfakes and misleading content, it will be a smooth ride for creative individuals and teams around the world using Sora in their work.