Film & Previsualization
Use Ray 3.2 text-to-video to prototype story beats, test camera angles, and build shot references before production begins. Generate cinematic sequences at 1080p HDR in minutes.
Powered by Ray 3.2 · Frame-Level AI Video
Ray 3.2 AI Video Generator
Direct any frame. Finish every cut.
Ray 3.2 packs four production workflows into one model — text-to-video, image-to-video, video editing, and reframing. Get 1080p HDR output with up to 16 keyframes. No GPU, no installs.
Native 1080p HDRUp to 16 Keyframes16-bit EXR ExportUp to 20s Clips
Available through leading creative platforms
- Adobe
- Freepik
- Fal.ai
- Comfy
- Lovart
- Envato
- Runware
- Weavy
Model Overview
What Is Ray 3.2?
Ray 3.2 is Luma AI's most capable video model — built for directors, studios, and creative teams who need frame-level control over every shot in their pipeline.
Luma Ray 3.2 is an AI video generation model that supports four production workflows: text-to-video, image-to-video, video editing (Modify Video V2), and video reframing. What sets Ray 3.2 apart from earlier Ray models is multi-keyframe control — you can place up to 16 keyframes inside a single clip and tell the model exactly where motion lands, where lighting shifts, and where each story beat resolves.
Ray 3.2 outputs native 1080p HDR and 16-bit EXR frames, which drop directly into color grading and compositing pipelines without upscaling. Whether you work in film, advertising, gaming, or social content, Ray 3.2 is designed to fit the tools your team already runs.
01
Frame-Level Control
Place up to 16 keyframes inside a single Ray 3.2 clip. Direct exactly where the look shifts, where a performance lands, and how each beat resolves — frame by frame, not clip by clip.
02
Native HDR + EXR Output
Ray 3.2 generates native 16-bit HDR color — not a post-process filter. Export as EXR in ACES2065-1 (AP0) color space, comp-ready for Nuke, DaVinci Resolve, and professional VFX pipelines.
03
Four Workflows in One Model
Generate from text, animate a still image, restyle existing footage with Ray 3.2 Modify Video V2, or reframe any clip to a new aspect ratio. One model covers the full production chain.
Ray 3.2 Workflows
Ray 3.2 Text to Video, Image to Video, Video Edit & Reframe
Ray 3.2 covers every stage of video creation — from generating new clips to editing footage you already have. Here's what each workflow does.
Text to Video
Ray 3.2 Text to Video — From Prompt to Cinematic Clip
Write a detailed scene description and Ray 3.2 generates a cinematic video clip with accurate motion, camera behavior, and lighting. Works for trailers, ads, social content, and creative prototyping. No source footage required.
5s or 10s clips540p / 720p / 1080pMultiple aspect ratiosOptional reference input
Image to Video
Ray 3.2 Image to Video — Animate Any Still Into Motion
Upload a reference image and describe the motion, camera path, and emotional tone you want. Ray 3.2 animates the image into a cinematic clip while preserving the visual identity of your source. Ideal for concept art, character shots, and illustrated scenes.
Up to 16 keyframes5s or 10sJPG / PNG / WEBP inputOptional reference
Video Edit
Ray 3.2 Video Edit — Restyle Footage, Keep the Performance
Upload a clip, describe the visual change you want, and Ray 3.2 transforms it while preserving the original motion, timing, and scene structure. Swap environments, change outfits, restyle lighting — the actor's performance stays intact. Supports up to 20 seconds at 1080p.
Up to 20s clips8-face performance trackingMotion-preservingMP4 / MOV / WMV input
Video Reframing
Ray 3.2 Video Reframing — One Clip for Every Platform
Change the aspect ratio of any clip without cropping. Ray 3.2 Reframe fills newly exposed canvas areas with scene-consistent, prompt-guided content that matches the original mood, lighting, and environment. Turn a 16:9 cinematic shot into a 9:16 vertical in minutes.
9:16 / 16:9 / 1:1 / 4:3 / 21:9Source up to 12sScene-aware fillPer-second pricing
Advanced Capabilities
Ray 3.2 Capabilities: 8 Production-Grade Features
Beyond the four core workflows, Ray 3.2 handles the most demanding visual production tasks.
Multi-Keyframe Control
Set up to 16 keyframes across a single clip. Direct the exact moments where motion shifts, lighting changes, or the story resolves.
Expressive Facial Performance
Transfer full facial expressions across shots. Tracks up to 8 faces simultaneously with skeletal pose and posture accuracy.
Motion Transfer
Extract motion from any source clip and apply it to new characters, objects, or scenes — preserving choreography and camera dynamics.
Character Transformation
Transform people into entirely new characters — different appearance, costume, species — while the performance carries through unchanged.
Environment Change
Swap the setting of any scene — location, season, lighting, world — without rebuilding the shot from scratch.
Relighting
Change the time of day, mood, or light source across an entire clip while keeping motion and subject intact.
Product Swap
Modify the product in an existing scene for different SKUs, seasonal variants, or market-specific campaigns — no reshoot needed.
Campaign Localization
Create market-specific video variants from one master clip — different settings, talent, or cultural context — at scale.
Technical Specs
Ray 3.2 Tech Specs: 1080p HDR, EXR & API
Ray 3.2 outputs drop cleanly into the timelines and compositing tools your team already uses.
1080p
Native 1080p Output
Real 1080p — no upscaling. Available across all four Ray 3.2 workflows.
HDR
Native HDR Generation
16-bit color generated by the model — not post-processed. Vivid, accurate, gradeable from the first render.
EXR
16-bit EXR Export
Export as EXR in ACES2065-1 (AP0) color space. Comp-ready for Nuke, DaVinci, and professional VFX pipelines.
20s
Extended Clip Length
Modify Video supports up to 20s at 24fps. T2V and I2V generate 5s or 10s native clips.
16 KF
Multi-Keyframe Support
Up to 16 keyframes per clip for both Modify Video and Image-to-Video workflows.
API
Full API Access
Ray 3.2 is available via the Luma API. Integrate any workflow directly into your tools, pipelines, and production software.
Who Uses Ray 3.2
Who Uses Ray 3.2?
Ray 3.2 works across industries where video quality and directorial control actually matter.
Advertising & Brand Video
Build product videos, mood-driven commercial concepts, and market-specific variants using Ray 3.2 video editing and campaign localization — without reshooting every SKU or market.
Game & Animation Concept
Animate concept art and character illustrations with Ray 3.2 image-to-video. Prototype cinematic cutscenes, creature animations, and environmental effects for pitching and pre-production.
Social Media Content
Generate short-form vertical clips, reframe existing footage with Ray 3.2 Video Reframing, and produce platform-native content for TikTok, Reels, and Shorts at speed.
VFX & Post Production
Export Ray 3.2 HDR renders as 16-bit EXR frames in ACES2065-1 color space — comp-ready for Nuke and DaVinci Resolve pipelines without additional conversion steps.
Agency & Studio Pipelines
Integrate Ray 3.2 via the Luma API into existing production tools and automate video generation workflows at scale for clients in entertainment, fashion, and e-commerce.
Ray 3.2 vs. Ray 3
Ray 3.2 vs Ray 3: Key Upgrades
Ray 3.2 builds directly on Ray 3 with meaningful production upgrades — not just performance tweaks.
| Feature | Ray 3.2 | Ray 3 / 3.14 | Earlier Models |
|---|---|---|---|
| Max Output Resolution | 1080p native | 1080p (3.14) | 720p / upscaled |
| Keyframe Control | Up to 16 keyframes | Start + end only | Start frame only |
| Modify Video (V2V) | V2 — up to 20s | V1 — shorter clips | Not available |
| Facial Performance Tracking | Up to 8 faces | Limited | Not supported |
| Native HDR + EXR Export | 16-bit, ACES AP0 | Yes (Ray 3) | No |
| Video Reframing | Custom positioning | Basic reframe | No |
| Text-to-Video | Yes | Yes | Yes |
| Image-to-Video | Yes | Yes | Yes |
Information from Luma AI official documentation as of June 2026. Subject to change with model updates.
Common Questions
Frequently Asked Questions
Everything you need to know before you generate your first clip.
Ray 3.2 is Luma AI's most advanced video generation model, released June 9, 2026. It powers four workflows — text-to-video, image-to-video, video editing (Modify Video V2), and video reframing — at up to native 1080p HDR with 16-bit EXR export. It's built for film, advertising, and professional creative production.
Ray 3.2 introduces three major upgrades: multi-keyframe control (up to 16 keyframes per clip instead of just start and end), upgraded Modify Video V2 with clips up to 20 seconds and Expressive Facial Performance tracking for up to 8 faces, and custom-positioning Video Reframe. Ray 3.14 improved speed and quality over Ray 3; Ray 3.2 adds directorial control.
Text-to-video and image-to-video: 360p (draft), 540p, 720p, or 1080p — at 5s or 10s. Modify Video (V2V): up to 20s at 1080p, depending on source frame rate (24fps = 20s max, 30fps = 15s, 60fps = 7s). Reframe: source video up to 12 seconds, output at 540p, 720p, or 1080p.
Ray 3.2 video-to-video (Modify Video V2) transforms an existing clip using a text prompt — changing style, environment, character look, or lighting — while preserving the original motion, timing, and performance. It tracks skeletal poses and up to 8 faces simultaneously. This makes it ideal for restyling footage without reshooting.
Earlier Ray models took a start frame and sometimes an end frame, generating everything in between automatically. Ray 3.2 lets you place up to 16 keyframes at specific moments in the clip. You export frames, edit them in tools like Photoshop or Nuke, and feed them back at the exact source-frame indexes — giving you precise control over every narrative beat.
Yes. Ray 3.2 text-to-video generates cinematic clips from natural-language prompts. Write a detailed scene description — including subject, motion, camera movement, lighting, and mood — and the model generates a 5s or 10s clip at your selected resolution. It also accepts optional reference inputs for additional visual guidance.
Commercial usage rights are governed by your Luma AI subscription agreement. This site is an independent third-party tool, not an official Luma product. Review Luma AI's terms of service at lumalabs.ai for full commercial licensing details before using generated content in paid campaigns or client deliverables.
Yes. Ray 3.2 is available through the Luma AI API. Developers, studios, and agencies can integrate all four workflows — text-to-video, image-to-video, video editing, and reframing — including multi-keyframe control, performance tracking, and EXR output, directly into their own tools and pipelines.