Ray 3.2

What is Ray 3.2?

Ray 3.2 is Luma AI's latest video generation model, designed to turn text prompts, still images, and existing footage into cinematic, frame-accurate clips. It operates entirely in the cloud, so you don't need a high-end local GPU to produce professional-grade results. The model supports four core workflows — text-to-video, image-to-video, video editing, and video reframing — all accessible directly from your browser.

What sets Ray 3.2 apart is its focus on frame-level control. You can place up to 16 keyframes within a single clip to direct lighting, camera movement, and character motion with precision. It natively outputs 1080p 16-bit HDR video and EXR sequences in ACES2065-1 color space, making it a practical tool for professional post-production pipelines.

Who it's for

• Video editors — who need to restyle existing footage while preserving the original performance, timing, and scene structure.
• Content marketers — who want to produce platform-specific variants (like 9:16 vertical clips) without reshooting or manually cropping.
• Creative directors and VFX artists — who require high-fidelity HDR output and EXR sequences compatible with Nuke, DaVinci Resolve, or other VFX pipelines.

Key features

Frame-level control with up to 16 keyframes

Ray 3.2 lets you place up to 16 keyframes inside a single clip. You can direct exactly where the look shifts, where a performance lands, and how each beat resolves — frame by frame, not clip by clip. This level of precision is rare in AI video tools and gives you granular creative authority.

Native 16-bit HDR and EXR export

The model generates real 16-bit HDR color — not a filter applied afterward. You can export as EXR in ACES2065-1 (AP0) color space, ready to drop straight into Nuke, DaVinci Resolve, or any VFX pipeline. This makes Ray 3.2 a viable option for professional post-production work.

Four workflows in one model

Ray 3.2 covers the full production chain: generate from text, animate a still image, restyle existing footage with the Modify Video V2 workflow, or reframe any clip to a new aspect ratio. One model handles everything from concept to final cut.

Motion-preserving video editing

When you upload a clip and describe a visual change, Ray 3.2 transforms it while keeping the original motion, timing, and scene structure intact. It supports up to 20-second clips at 1080p and tracks up to eight faces, so performances stay consistent even when environments, outfits, or lighting change.

What stands out

"Direct any frame. Finish every cut."

Ray 3.2's edge is its combination of frame-level precision and professional-grade output. Most AI video tools give you a clip and hope for the best; Ray 3.2 lets you steer the result with keyframes, reframing, and scene-aware fills. The native HDR and EXR support means the output isn't just social-media-ready — it's pipeline-ready. For teams that need campaign variants, concept visualizations, or VFX plates without a full production budget, this is a practical, scalable starting point.

Worth checking out if…

You're a creator or studio that needs cinematic video with real control — whether you're restyling existing footage, animating stills, or reframing clips for different platforms. If you work in post-production and require HDR or EXR output, or if you want to produce multiple campaign variants from a single source clip, Ray 3.2 is worth a serious look. It's also a strong fit if you want professional results without investing in expensive local hardware.