Ray 3.2

Ray 3.2 is a production-grade video generation and transformation model developed by Luma AI. Designed for professional production environments in entertainment, advertising, and gaming, Ray 3.2 emphasizes frame-level control and creative direction over purely random generative outputs, allowing creators to bridge text descriptions, images, and source footage into consistent cinematic sequences.

Key Capabilities and Workflows

The model supports multiple core workflows, including text-to-video, image-to-video, video-to-video transformation, and reframing. A primary capability is its multi-keyframe support, which lets users place up to 16 keyframes within a single clip to dictate motion, maintain character consistency, and guide complex story beats. Clips can be generated at 5 or 10 seconds in durations across resolutions up to 1080p and various aspect ratios.

Professional Integration

Ray 3.2 incorporates native HDR generation and 16-bit EXR export using the ACES2065-1 standard, aligning with professional color-grading and VFX post-production pipelines. Operating strictly as a visual generation asset, it requires separate audio integration while providing studios with structured control over look development and visual style transformation.