Sora, an artificial intelligence (AI) video generator program created by startup OpenAI in 2021, is making waves as it has now moved out of the research phase and has been officially released to the public under the new name of Sora Turbo.
The launch has triggered an online frenzy among users, causing the company to temporarily halt new account creations after finding itself overwhelmed by an avalanche of traffic.
Sora uses text prompts to create content, similar to other content creation programs such as ChatGPT. Unlike traditional AI programs which produce written responses, Sora creates high-quality videos based on a user’s text input.
Sora is disgustingly good, gj @sama
Adding it to @everartai asap so you can bring your images to life pic.twitter.com/wMehxOc8cm
— Jay A (@jay_azhang) December 9, 2024
According to a statement from OpenAI which accompanied the release on December 9, Sora Turbo is an improvement on the original Sora program as it allows for faster video creation, higher quality videos and a host of video editing features which enable users to clip and create video sequences, and to change elements in a video while preserving other parts of the video.
So what exactly is Sora Turbo and how is it changing the video creation industry?
How does Sora Turbo work?
The power of Sora, and many other AI video generators, is its ability to replicate what humans can do with high-quality film equipment but in an extremely short time.
Users can give the program prompts not just on what to include in the video, but also on which type of camera they want to use, lens selection, lighting setup, visual aesthetics, composition and storyboarding.
An example prompt using some of these details could be as follows:
“Use a medium-telephoto lens (85mm) to capture an intimate portrait of a street vendor working at night. Illuminate the scene with a mix of the vendor’s own warm, glowing stall lights and cooler ambient city lights in the background. Focus on the details of the vendor’s hands as they prepare food or goods, balancing the frame with their expressive face. The image should evoke the hard work and pride in their craft.”
How good is Sora Turbo?
It is not perfect.
Users write a text prompt about what type of video they would like to generate and Sora creates the video. There are limitations in the current release. Video quality is up to 1080p at 30 frames per second and can only be 20 seconds long.
This is an example of a Sora Turbo video from Emi Kusana, a multimedia artist and the lead singer of Satellite Young, a music group inspired by 1980s J-pop and sci-fi themes, using themes from Takenoko-zoku, a fashion group from the 1980s. Although there is a 20-second limit, the Sora Storyboard feature allows users to string multiple AI-generated videos together and arrange them sequentially, as shown in the video below.
Obvious errors also appear in some videos. For example, in the above video, at the 36-second mark, one of the dancers does a 180-degree neck turn, which of course is not humanly possible, and continues the dance routine. The company will have to work further to resolve bugs like this, it said.
In an example video of a woman cooking, posted in February by Sam Altman, founder of OpenAI, viewers will spot that the spoon in the woman’s right hand disappears after she stops mixing.
https://t.co/rmk9zI0oqO pic.twitter.com/WanFKOzdIw
— Sam Altman (@sama) February 15, 2024
How has Sora Turbo been received within the industry?
Some industry experts and creative directors are worried about how Sora may disrupt the film industry, particularly Hollywood.
In March, one Hollywood executive told Al Jazeera on condition of anonymity: “I don’t see it as a threat to production so much as a threat to the way production is done as we currently know it.”
He referred to changes in the way post-production was carried out in the past – such as when people began editing video on personal laptops rather than paying post-production companies to do the work.
“Lots of people got wiped out in that transition while others could suddenly afford a proper editor without the overhead a post house demands,” he said.
Is Sora the only AI video generator?
No, it is not. The market for AI video generators is growing every day. One clear differentiator between Sora and other competitors, is the unmatched video quality and realism that Sora is able to produce, however.
Moreover, the new release, Sora Turbo, comes with a host of advanced video editing features that users might find on Adobe Premiere Pro or Final Cut Pro, two very popular video editing software programs used in film, TV and online content creation.
In addition, Sora is financially backed by mega software company Microsoft. To date, Microsoft has invested almost $14bn with the latest cash injection of $10bn on January 23. According to CNBC in October, OpenAI now has a $4bn revolving line of credit, bringing its total liquidity to more than $10bn. That amount of investment provides Sora with a level of leverage which many other competitors don’t have.
Which other video generators are available?
There are several other top-tier AI video generators, including:
Runway Gen 3
Runway, founded in 2018 has raised a total of $237m in funding and is reportedly in talks with venture capital firm General Atlantic to raise $450m in new funding, according to The Information, a publication that covers the technology industry and venture capital industry. The free version of Runway Gen 3 can generate 10-second video clips at 720p resolution with 30 frames per second. For paid subscribers, video clips can be set to 1440p or 2160p.
Similar to Sora, it can create a video based on a text prompt but leans heavily on social media creators and marketers who use it to create advertising due to its high speed and efficiency in short-form video projects.
One of the challenges of long-form AI-generated video is consistency across multiple generated videos. Longer videos require interactive prompt refinement to make all videos look and feel the same.
Kling AI
Kling, backed by Kuaishou, which is also known internationally as Kwai, is a major Chinese technology company based in Beijing, China. Kling was unveiled in June 2024 and is positioning itself as a competitor to platforms like OpenAI’s Sora. Kling can generate videos up to two minutes long with a resolution of 1080p at 30 frames per second.
MiniMax
MiniMax, a Chinese AI startup is backed by Alibaba Group Holding, a multinational company specialising in e-commerce, internet services and technology services based in China and whose most recent round of funding was $600m in March. MiniMax was founded in 2021 by former employees of SenseTime, a leading AI company based in Shatin, Hong Kong.
Minimax can generate six-second video clips at 720p resolution and 25 frames per second.
