Be a part of our each day and weekly newsletters for the most recent updates and unique content material on industry-leading AI protection. Study Extra
Runway AI Inc. launched its most superior AI video technology mannequin right this moment, coming into the following part of competitors to create instruments that might remodel movie manufacturing. The brand new Gen-4 system introduces character and scene consistency throughout a number of photographs — a functionality that has evaded most AI video turbines till now.
The New York-based startup, backed by Google, Nvidia and Salesforce, is releasing “Gen-4” to all paid subscribers and enterprise prospects, with further options deliberate for later this week. Customers can generate 5 and ten-second clips at 720p decision.
The discharge comes simply days after OpenAI’s picture technology characteristic created a cultural phenomenon, with hundreds of thousands of customers requesting Studio Ghibli-style photos by ChatGPT. The viral development turned so in style that it quickly crashed OpenAI’s servers, with CEO Sam Altman tweeting that “our GPUs are melting” as a consequence of unprecedented demand. The Ghibli-style photos additionally sparked heated debates about copyright, with many questioning whether or not AI firms can legally mimic distinctive creative types.
Visible continuity: The lacking piece in AI filmmaking till now
Character and scene consistency — sustaining the identical visible parts throughout a number of photographs and angles — has been the Achilles’ heel of AI video technology. When a personality’s face subtly modifications between cuts or a background component disappears with out clarification, the substitute nature of the content material turns into instantly obvious to viewers.
The problem stems from how these fashions work at a elementary stage. Earlier AI turbines handled every body as a separate artistic job, with solely unfastened connections between them. Think about asking a room stuffed with artists to every draw one body of a movie with out seeing what got here earlier than or after — the outcome can be visually disjointed.
Runway’s Gen-4 seems to have tackled this downside by creating what quantities to a persistent reminiscence of visible parts. As soon as a personality, object, or setting is established, the system can render it from totally different angles whereas sustaining its core attributes. This isn’t only a technical enchancment; it’s the distinction between creating attention-grabbing visible snippets and telling precise tales.
Utilizing visible references, mixed with directions, Gen-4 permits you to create new photos and movies with constant types, topics, places and extra. Permitting for continuity and management inside your tales.
To check the mannequin’s narrative capabilities, we’ve got put collectively… pic.twitter.com/IYz2BaeW2U
— Runway (@runwayml) March 31, 2025
In line with Runway’s documentation, Gen-4 permits customers to supply reference photos of topics and describe the composition they need, with the AI producing constant outputs from totally different angles. The corporate claims the mannequin can render movies with lifelike movement whereas sustaining topic, object, and elegance consistency.
To showcase the mannequin’s capabilities, Runway launched a number of quick movies created totally with Gen-4. One movie, “New York is a Zoo,” demonstrates the mannequin’s visible results by inserting lifelike animals in cinematic New York settings. One other, titled “The Retrieval,” follows explorers looking for a mysterious flower and was produced in lower than every week.
From facial animation to world fashions: Runway’s AI filmmaking evolution
Gen-4 builds on Runway’s earlier instruments. In October, the corporate launched Act-One, a characteristic that permits filmmakers to seize facial expressions from smartphone video and switch them to AI-generated characters. The next month, Runway added superior 3D-like digicam controls to its Gen-3 Alpha Turbo mannequin, enabling customers to zoom out and in of scenes whereas preserving character varieties.
This trajectory reveals Runway’s strategic imaginative and prescient. Whereas rivals deal with creating ever extra lifelike single photos or clips, Runway has been assembling the parts of a whole digital manufacturing pipeline. The method feels extra akin to how precise filmmakers work — addressing issues of efficiency, protection, and visible continuity as interconnected challenges reasonably than remoted technical hurdles.
The evolution from facial animation instruments to constant world fashions suggests Runway understands that AI-assisted filmmaking must observe the logic of conventional manufacturing to be really helpful. It’s the distinction between making a tech demo and constructing instruments professionals can truly incorporate into their workflows.
AI video’s billion-dollar battle heats up
The monetary implications are substantial for Runway, which is reportedly elevating a new funding spherical that may worth the corporate at $4 billion. In line with monetary experiences, the startup goals to achieve $300 million in annualized income this 12 months following the launch of recent merchandise and an API for its video-generating fashions.
Runway has pursued Hollywood partnerships, securing a take care of Lionsgate to create a customized AI video technology mannequin primarily based on the studio’s catalog of greater than 20,000 titles. The corporate has additionally established the Hundred Movie Fund, providing filmmakers as much as $1 million to supply films utilizing AI.
“We believe that the best stories are yet to be told, but that traditional funding mechanisms often overlook new and emerging visions within the larger industry ecosystem,” Runway explains on its fund’s web site.
Nevertheless, the expertise raises issues for movie {industry} professionals. A 2024 examine commissioned by the Animation Guild discovered that 75% of movie manufacturing firms which have adopted AI have diminished, consolidated, or eradicated jobs. The examine initiatives that greater than 100,000 U.S. leisure jobs can be affected by generative AI by 2026.
Copyright questions observe AI’s artistic explosion
Like different AI firms, Runway faces authorized scrutiny over its coaching information. The corporate is at the moment defending itself in a lawsuit introduced by artists who allege their copyrighted work was used to coach AI fashions with out permission. Runway has cited the honest use doctrine as its protection, although courts have but to definitively rule on this software of copyright legislation.
The copyright debate intensified final week with OpenAI’s Studio Ghibli characteristic, which allowed customers to generate photos within the distinctive model of Hayao Miyazaki’s animation studio with out specific permission. Not like OpenAI, which refuses to generate photos within the model of residing artists however permits studio types, Runway has not publicly detailed its insurance policies on model mimicry.
This distinction feels more and more arbitrary as AI fashions turn out to be extra subtle. The road between studying from broad creative traditions and copying particular creators’ types has blurred to close invisibility. When an AI can completely mimic the visible language that took Miyazaki many years to develop, does it matter whether or not we’re asking it to repeat the studio or the artist himself?
When questioned about coaching information sources, Runway has declined to supply specifics, citing aggressive issues. This opacity has turn out to be normal apply amongst AI builders however stays some extent of competition for creators.
As advertising and marketing businesses, instructional content material creators, and company communications groups discover how instruments like Gen-4 might streamline video manufacturing, the query shifts from technical capabilities to artistic software.
For filmmakers, the expertise represents each alternative and disruption. Impartial creators acquire entry to visible results capabilities beforehand out there solely to main studios, whereas conventional VFX and animation professionals face an unsure future.
The uncomfortable fact is that technical limitations have by no means been what prevents most individuals from making compelling movies. The flexibility to keep up visible continuity received’t all of a sudden create a technology of storytelling geniuses. What it’d do, nevertheless, is take away sufficient friction from the method that extra individuals can experiment with visible narrative while not having specialised coaching or costly tools.
Maybe essentially the most profound facet of Gen-4 isn’t what it may possibly create, however what it suggests about our relationship with visible media going ahead. We’re coming into an period the place the bottleneck in manufacturing isn’t technical talent or funds, however creativeness and objective. In a world the place anybody can create any picture they will describe, the vital query turns into: what’s price exhibiting?
As we enter an period the place creating a movie requires little greater than a reference picture and a immediate, essentially the most urgent query isn’t whether or not AI could make compelling movies, however whether or not we are able to discover one thing significant to say when the instruments to say something are at our fingertips.
