Luma releases a new AI model that lets users generate a video from a start and end frame (techcrunch.com)

Luma, an a16z-backed AI video and 3D modeling company, has unveiled its latest model, Ray3 Modify, which empowers users to generate videos from specified start and end frames while retaining the original performance of human actors. This innovative model addresses a common challenge in AI video editing by allowing creators to modify existing footage based on character reference images, ensuring that key aspects such as movement, timing, and emotional delivery remain intact. This capability enables creative studios to change an actor's appearance and costume without sacrificing the authenticity of their performance. Significantly, Ray3 Modify offers unprecedented control for creative teams, allowing them to capture performances with a standard camera and instantly alter locations, outfits, or even reshoot scenes using AI, all while maintaining continuity. Released through Luma's Dream Machine platform, this tool positions the company competitively against rivals like Runway and Kling. The launch follows a substantial $900 million funding round led by a Public Investment Fund-owned AI company, highlighting investor confidence in Luma's potential to reshape the creative landscape with its advanced AI technologies.