On Thursday OpenAI released Sora, their text-to-video model. It is remarkable, and I plan to write more about it and the future of video more generally in a forthcoming article.

Here’s an example of what it can do…

Prompt:

A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick. She walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about.

Output:

This video was generated completely by the AI model. No 3D rendering or other inputs. It is, simply put, lightyears ahead of what we’ve seen from the likes of Google’s Lumiere and Runway’s Gen-2 which were the state of the art.

The most impressive details aren’t on the main Sora page though, they’re hidden away in the technical report. Not only can Sora generate entirely fresh video as shown above, but it can also transform existing videos significantly, based on a text prompt:

There are obviously huge implications to this for video entertainment, but one of the more fascinating aspects of this for me is in what it means for ad creative.

Yes, we might not like video ads but they are sadly here to stay. One of the reasons we don’t like them (beyond the interruption) is that they often feel so un-targeted and repetitive. That’s not entirely surprising given that the cost of making a 30-second commercial runs into the 10s of thousands of dollars let alone the actual time it takes to produce.

From the preview that we saw on Thursday though, these videos can be made in less than 10 minutes and at significantly reduced cost. Whilst the prospect of seeing hyper-targeted videos akin to last decade’s oddly specific t-shirt ads is not appealing, from a brand perspective being able to quickly and cheaply produce not only a series of ads targeted at each and every demographic but also alternatives to A/B test with is a marketer’s dream.

Film one video and then have AI produce modified versions for different ages, locales and cultural norms – a little dystopian? Probably. Inevitable? Absolutely.

Sure, we’re (maybe) unlikely to see this footage used in the Super Bowl, but as YouTube ads? Absolutely. And really, I’m not entirely averse to receiving YouTube ads that fit with my interests as well as Instagram ads tend to.

It’s not just preplanned creative that will benefit here either – social media teams seeking to produce responsive ads that match the calibre of Oreo’s famous 2013 tweet in response to a power cut at the Super Bowl now have a tool that can produce high-production quality video in minutes.