Runway, a New York City-based AI startup, has released its latest AI video generator, Gen-3 Alpha, and judging by the sample clips the company has shared so far, it's pretty impressive.
From an astronaut running through the alleys of Rio de Janeiro to a realistic suburban neighborhood submerged in water and surrounded by coral reefs, Gen-3 Alpha is a reminder of just how far generative AI has come.
The level of fidelity is impressive, from strange, menacing creatures roaming streetlight-lit streets to a woman running toward a rocket launching. Human faces in particular are incredibly realistic; one clip shows a bald man “suddenly having his curly wig and sunglasses drop onto his head.”
In short, it's a great glimpse into the horrors of the near future. And judging by the video, Runway's latest AI model holds its own against OpenAI's recently announced Sora (which isn't publicly available yet). But we'll wait until we have a chance to try out one of these tools first-hand before we call it a winner.
But it seems some online experts have already made up their minds.
“Even if it's handpicked, it's already better looking than Sora,” one Reddit user argued.
In addition to its latest video generator, Runway has also released a set of tweaking tools, including advanced camera controls.
The company claims that Gen-3 is a step towards a more ambitious goal of enabling a “general world model” — “an AI system that builds an internal representation of an environment and uses it to simulate future events within that environment.”
Like OpenAI's Sora, Runway has not yet announced an exact release date for the model, nor is it clear whether Runway will charge users for access. The company already sells subscriptions to its existing AI tools, including the two predecessors to Gen-3, as well as other AI-based video editing tools.
According to Runway co-founder and CTO Anastasis Germanidis, Gen-3 Alpha “will be available in Runway products shortly and will support all the existing modes you're used to using (text-to-video, image-to-video, video-to-video) as well as some new modes that were previously only possible on the more capable base model.”
Like OpenAI's AI video generator, the samples produced by Gen-3 Alpha are far from pixel perfect, with corrupted text and missing body parts.
Runway claims on its website that Gen-3 Alpha was “trained using a combination of video and imagery,” but doesn't elaborate on where this data comes from, a trend that's becoming increasingly common among AI companies.
Seemingly preempting concerns about possible copyright infringement, Runway also claims to be “working and partnering with major entertainment and media organizations to create custom versions of the Gen-3.”
We'll reserve judgment until we get a chance to test drive the Gen-3 Alpha, but so far, Runway is no joke.
More about AI Video: This AI-generated Pixar-esque animation might blow your mind
