OpenAI's Sora: Filmmakers say its 'surreal' edge is the real draw

OpenAI has released a new batch of videos created by early-access artists using Sora, the video generation model it dropped on an unsuspecting world last month. The company is framing this not as a polished product launch, but as a research collaboration with filmmakers, designers, and creative directors to see how the tool actually fits into a working creative process. The short answer, based on these seven case studies? The artists aren’t just using it to replicate reality. They’re drawn to its weirdness.

Toronto-based multimedia collective shy kids used Sora to make a short film about a man with a balloon for a head, titled “Air Head.” Director Walter Woodman captured the sentiment echoed by nearly every creator in the batch: “As great as Sora is at generating things that appear real, what excites us is its ability to make things that are totally surreal. A new era of abstract expressionism.” For a team that has long envisioned stories they lacked the budget to shoot, the tool represents a door opening. “People from all over the world with stories ready to burst out of their chests finally have the opportunity to show the world what’s inside,” Woodman added.

The other collaborators are singing from the same sheet music, but each highlights a different pain point Sora eliminates. Director Paul Trillo, a 19-time Vimeo Staff Pick honoree, described feeling “unchained” for the first time, free from the tyranny of time, money, and permissions. Native Foreign’s Nik Kleverov, an Emmy-nominated creative director, is using the model for rapid brand prototyping, noting that budgetary handcuffs no longer have to strangle a creative narrative. Digital artist Don Allen Stevenson III, who started his career at DreamWorks, zeroes in on the model’s physics-defying glitches as its killer feature: “It’s not bound by traditional laws of physics or conventions of thought.” He says it lets him prototype bizarre augmented-reality creatures before committing to a full 3D build.

It’s a remarkably unified front of artistic enthusiasm, which probably shouldn’t surprise anyone. These are hand-picked creators given early access to a tool that, by their own admission, lets them skip the grunt work. The real test isn’t in these polished highlight reels — each video was edited by the artists, who had full freedom to modify Sora’s output. The test will be how the tool behaves in the hands of the rest of us, when the novelty wears off and the technical constraints that Josephine Miller alluded to are still there, just shifted to a different part of the pipeline. For now, OpenAI is wise to let the artists do the talking.