Watch These Surreal AI Shorts By Indie Creators Using OpenAI Sora

shykids airhead sora
When we originally wrote about OpenAI's Sora back in the middle of February, we noted that the AI video model would be used by some independent creators before it became available to the public. The idea is that OpenAI wanted to assess whether the model had any real value before opening it up to the public, likely because it is quite computationally expensive to run. That's just our conjecture, but it makes sense.

Today, OpenAI revealed the fruits of some of those partnerships. Independent filmmakers, writer/directors, creative agencies, and digital artists have all revealed what they have created with Sora, and some of the results are honestly breathtaking. We're not completely being hyperbolic here; I literally gasped when watching one of the clips available on the "Sora First Impressions" page.

The individuals and groups involved in the "Sora First Impressions" page include Toronto-based multimedia production company "shy kids", writer/director Paul Trillo, the creative director for Native Foreign agency Nik Kleverov, musician August Kamp, creative director for London-based Orarr Studio Josephine Miller, former DreamWorks artist Don Allen Stevenson III, and OpenAI's own artist in residence, Alex Reben.

sora girafflamingo
Top: "Air Head" by shy kids, Above: "Beyond Our Reality" by Don Allen Stevenson III

If you watch only one of the clips, scroll down to the last one by Alex Reben. It's short, only 20 seconds, but it shows off some impressive capabilities for AI video, including the ability to rotate an object in 3D space and have it remain consistent in size and form. Normally AI video and image generators have no real concept of what an object looks like in 3D space, and only work in the 2D space. It's a cool demo.

Many of the shorts focus on the surreal, because if you wanted to create something mundane, you could simply do that with normal video tools. Sora, as an AI, lets you create things that don't and couldn't exist. Most of the comments from the creators focus on this idea, but there's also the salient point, brought up by the shy kids production company, that generative AI gives people with no artistic talent the ability to express the ideas and stories in their heads in a way that wasn't possible before.

ai stable diffusion art2
Could you create this without AI? (generated using a blended Stable Diffusion model.)

Indeed, this is an argument we've made before. Generative AI can't actually replace artists; every single clip on the Sora AI site has visible flaws and obvious defects if you take a second to look for them. A filmmaker working with a camera and actors will always create something that is more natural and generally superior to anything you could produce with AI. The fact that someone with no artistic talent whatsoever can type some words and produce something that's interesting to watch is incredible, though.