A new interview with the director behind the viral Sora clip Air head has revealed that AI played a smaller role in production than initially claimed.
Revealed by Patrick Cederberg (who did post-production for the viral video) in a interview with Fxguide, it has now been confirmed that OpenAI’s text-to-video program was far from the only force involved in its production. The 1 minute and 21 second clip was created using a combination of traditional film techniques and post-production editing to achieve the look of the final photo.
Air head was created by ShyKids and tells the short story of a man with a literal balloon for a head. Although it uses a human voiceover, the way OpenAI pushed the clip on social channels like YouTube certainly left the impression that the footage was purely AI-powered, but that’s not entirely true.
As shown in the behind the scenes video, a ton of work was done by ShyKids, who took Sora’s raw output and helped clean it up into the final product. This included manually rotoscoping the backgrounds, removing the faces that occasionally appeared on the balloons, and color correcting.
Then there’s the fact that Sora needs a lot of time to get everything right. Cederberg explains that there were “hundreds of generations of 10 to 20 seconds each,” which were then tightly edited in what the team described as a “300:1” ratio of what was generated versus what was ready for further adjustments.
Such manual work also included editing the head that would appear and reappear, and even changing the color of the balloon itself, which would look red instead of yellow. While Sora was used to generating the first footage with good results, there was clearly a lot more going on behind the scenes to make the final product look as good as it does, so we’re still a long way from directly generated movies. quality productions.
Sora remains closely monitored, save for a handful of carefully curated projects that have surfaced Air head one of the most popular. The clip has been viewed more than 120,000 times at the time of writing, with OpenAI being touted as “experimenting” with the program, downplaying the obvious work that went into the final product.
Sora is impressive, but we’re not convinced
While OpenAI has done a good job of showing what its text-to-video service can do through the large language model, the lack of transparency is concerning.
Air head is an impressive clip from a talented team, but it took a lot of editing to get the final product where it is in the short film.
It’s not quite the one-click-and-you’re-done approach that many of the technology’s boosters have portrayed it as. It turns out that it’s just a tool that can be used to enhance footage rather than create something from scratch, something that’s already common enough in video production that it makes Sora seem less revolutionary than it initially seemed.