
Luma has released a new AI model called Ray3 Modify that allows creators to alter existing video footage while preserving the original human performance, the company said Thursday. The model enables users to modify scenes using character reference images and guided start and end frames, giving creative teams more control over AI-generated video edits without reshooting physical footage.
Preserving Human Performance In AI Video Editing
Luma said Ray3 Modify addresses a long-standing challenge for creative studios using generative video tools: maintaining authentic human performance while applying visual changes. According to the company, the model follows the input footage more closely than previous approaches, allowing studios to continue working with human actors for branded or creative projects.
The company said the model retains the actor’s original motion, timing, eye line, and emotional delivery while changing the surrounding scene or visual elements. This allows edits and effects to be applied without altering how the performance itself appears.
Character References And Visual Continuity
With Ray3 Modify, users can upload character reference images to transform the appearance of a human actor in the original footage. Luma said these references help preserve details such as costumes, likeness, and identity throughout a shoot, even as scenes are modified.
The company said this approach allows creators to maintain visual consistency across multiple shots while applying AI-based changes to characters or environments.
Guided Transitions Using Start And End Frames
Ray3 Modify also allows users to define a start frame and an end frame to guide how the model generates transitional footage. Luma said this feature helps creators direct scene transitions, control character movement or behavior, and maintain continuity between shots.
The company said this capability gives creative teams more precise control over how AI-generated footage evolves from one moment to the next.
Company Statement On Creative Control
Amit Jain, co-founder and CEO of Luma AI, said generative video tools offer strong creative expression but are often difficult to control. In a statement, Jain said Ray3 Modify combines real-world footage with AI-generated changes while giving creators more direct control over the final output.
Jain said the model allows teams to capture performances on camera and then modify locations, costumes, or scenes using AI, including revisiting shots digitally without recreating the original physical production.
Availability And Competitive Landscape
Luma said Ray3 Modify is available through its Dream Machine platform. The company operates in a competitive space alongside video generation and editing companies such as Runway and Kling. Luma previously released video modification features in June 2025.
Funding And Infrastructure Plans
The release follows a $900 million funding round announced in November. The round was led by Humain, an AI company owned by Saudi Arabia’s Public Investment Fund. Existing investors, including a16z, Amplify Partners, and Matrix Partners, also participated.
Luma said it is planning to build a 2GW AI cluster in Saudi Arabia in partnership with Humain as part of its infrastructure expansion.
Featured image credits: Luma AI
For more stories like it, click the +Follow button at the top of this page to follow us.
