AI video startup Runway has announced a new, more advanced AI video generation model, which the company is calling Gen-4. With this model, Runway hopes to solve several of the most difficult challenges that plague AI video generation. The company’s announcement describes its Gen-4 model as “a new generation of consistent and controllable media.”
Addressing challenges in AI video generation
One of the chief difficulties that the new model seeks to overcome is the dream-like, often incoherent nature of AI-generated videos. Current models struggle to maintain consistent characters and objects across frames, so their appearance often changes drastically depending on the environment and lighting conditions.
Runway has released multiple examples of videos to demonstrate how the Gen-4 AI model can solve these issues. For instance, one of these videos showed the same statue in different locations, and the object looked consistent despite the changes in camera angles and lighting conditions. This level of spatial consistency — including the ability to render multiple angles of the same scene — is a new ability exclusive to Gen-4. Earlier versions, Gen-2 and Gen-3, lacked this functionality.
“With Gen-4, you are now able to precisely generate consistent characters, locations and objects across scenes,” says the new model’s landing page. “Simply set your look and feel and the model will maintain coherent world environments while preserving the distinctive style, mood and cinematographic elements of each frame. Then, regenerate those elements from multiple perspectives and positions within your scenes.”
The new AI generation tool is rolling out to paid Runway users in phases starting on Monday. Paid users who don’t currently see that Gen-4 option in their account should check back later.
Runaway carves out niches AI video generation
Runway may not be a household name like other major AI companies, such as OpenAI, which have vast amounts of funding to spend on both development and marketing. However, Runway is carving out a space for itself as a quality-first AI video generator that appeals to filmmakers, designers, and other creators.
Notably, Runway was among the first companies to launch a workable AI video generation tool for public use, back in early 2023. Subsequent versions have been employed in real-world creative projects, albeit in a limited way. Both The Late Show with Stephen Colbert and the film Everything Everywhere All at Once have used the AI tool to generate animation sequences. Viewers may see more visuals created with Runway AI now that the new model has launched.