Runway, the US-based AI startup, has taken another significant step in the rapidly evolving field of AI-generated video. The company announced today that its Gen-3 Alpha Image to Video tool now supports using an image as either the first or last frame of video generation, a feature that could dramatically improve creative control for filmmakers, marketers, and content creators.
The startup was founded in 2018 by Cristóbal Valenzuela, Alejandro Matamala, and Anastasis Germanidis.
Furthermore, this update comes after the startup officially released Gen-3 Alpha, highlighting the company’s aggressive push to stay ahead in the competitive AI video generation market.
The new capabilities of the model allows users to anchor their AI-generated videos with specific imagery, potentially solving one of the key challenges in AI video creation; consistency and predictability.
By allowing users to generate high-quality, ultra-realistic scenes that are up to 10 seconds long—with various camera movements—using only text prompts, still imagery, or pre-recorded footage, this model has set a new benchmark in video creation.
“The ability to create unusual transitions has been one of the most fun and surprising ways we’ve been using Gen-3 Alpha internally,” said Runway co-founder and CTO Anastasis Germanidis.
Back in February 2023, Runway released Gen-1 and Gen-2, the first commercial and publicly available foundational video-to-video and text-to-video generation models accessible via an easy-to-use website. Now the Gen-3 update takes it to the next level.
The Power of First and Last Frames
“Gen-3 Alpha update now supports using an image as either the first or last frame of your video generation. This feature can be used on its own or combined with a text prompt for additional guidance,” Runway announced on X.
The impact of this feature was immediately recognised by users. Justin Ryan, a digital artist, posted in response: “This is such a big deal! I’m hoping this means we are closer to the First and final frame like Luma Labs offers.”
This development puts Runway in direct competition with other players in the space, such as Luma Labs, Pika, OpenAI’s much-anticipated Sora, and the Bengaluru-based startup Unscript, which is generating videos using single images.
However, Runway’s public availability gives it a significant edge over Sora, which remains in closed testing.
A spokesperson from Runway shared that the initial rollout will support 5 and 10-second video generations, with significantly faster processing times. Specifically, a 5-second clip will take 45 seconds to generate, while a 10-second clip will take 90 seconds.
Accelerating to Get Ahead
Since the release of the Gen-3 Alpha model, internet users have been showcasing their unique creations in high-definition videos, demonstrating the versatility and range of Runway AI’s latest AI model.
As Runway makes a bold move, there is a significant shift in the generative AI video space. The company describes this update as “first in a series of models developed by Runway on a new infrastructure designed for large-scale multimodal training,” and a “step toward creating General World Models.”
Germanidis also revealed that Gen-3 Alpha will soon enhance all existing Runway modes and introduce new features with its advanced base model.
He also noted that since Gen-2’s 2023 release, Runway has found that video diffusion models still have significant performance potential and create powerful visual representations.
While the startup states that Gen-3 Alpha was “trained on new infrastructure” and developed collaboratively by a team of researchers, engineers, and artists, it has not disclosed specific datasets, following the trend of other leading AI media generators that keep details about data sources and licensing confidential.
Interestingly, the company also notes that it has already been “collaborating and partnering with leading entertainment and media organisations to create custom versions of Gen-3,” which “allows for more stylistically controlled and consistent characters, and targets specific artistic and narrative requirements, among other features.”
AI comes to filmmaking
Additionally, Runway hosted its second annual AI Film Festival in Los Angeles. To illustrate the event’s growth since its inaugural year, Valenzuela noted that while 300 videos were submitted for consideration last year, this year they sent in 3,000.
Hundreds of filmmakers, tech enthusiasts, artists, venture capitalists, and notable figures, including Poker Face star Natasha Lyonne, gathered to watch the 10 finalists selected by the festival’s judges.
Now, the films look different, as does the industry with generative AI.
Meanwhile, amidst all this, it is evident that Runway is not giving up the fight to be a dominant player or leader in the rapidly advancing generative AI video creation space.
