Ray3.2: FAQ
What Is Ray3.2?
A: Ray3.2 is our newest and most versatile video model. It powers four workflows in one: Modify Video, image2video, and Reframe.
What’s new versus prior Ray models?
A: Keyframe control over the whole clip, not just its endpoints. Earlier Ray models took a single start frame (and sometimes an end frame) and generated everything in between on their own. Ray3.2 lets you set up to 64 keyframes, so you decide exactly where the look shifts, where a movement lands, and where a beat hits.
What features are available with Ray3.2?
A: Modify Video; Reframe; image2video; HDR generation (Native 16-bit color with EXR output)
What output formats are available with Ray3.2?
A: Standard outputs are generated in .mp4 format. In addition, Ray3.2 supports native 16-bit HDR generation and EXR export. These formats provide the wide gamut and color accuracy required for professional color grading, compositing, and VFX pipelines.
What resolution(s) and lengths are available for Ray3.2?
A: You can generate in 360p, 540p, 720p (default), and 1080p. Maximum length depends on the mode: text2video and image2video produce 5 or 10-second native clips, Modify Video (V2V) can provide up to 20 seconds video depending on frame rate, and Reframe supports source videos up to 16 seconds.
Do generations have native audio?
A: text2video and image2video: No, there's no natively-generated audio, so voiceover, music, and sound effects are added as a separate step.
Modify Video and Reframe will preserve your original audio. However, Audio generation is available in the app via ElevenLabs, included as part of your current plan.
Ray3.2 Modify Video
What is new in Modify Video?
A: Modify Video3.2 features multi-keyframe support, enhanced performance tracking for skeletal poses, and expressive facial capture. It also introduces 1080p native outputs with generation lengths extending up to 20 seconds (depending on source video Frame Rate).
What are the key new and enhanced capabilities in Ray3.2 Modify Video?
A: Key capabilities include:
Multi-Keyframe: Add up to 64 individual keyframes within a single clip for precise storybeat control.
Expressive Facial Performance: Transfers the full expression state, for up to 8 faces.
Performance Tracking: Holds skeletal pose, body, gesture, and posture with high accuracy across the clip.
Improved Motion Transfer: Extract or preserve motion from a source video and apply it to a new scene
How does my source video frame rate affect the length of my video?
A: The lower the frame rate, the longer the clip you can generate. Maximum length at each frame rate:
24fps: up to 20 seconds
30fps: up to 15 seconds
60fps: up to 7 seconds
What footage works best with Modify Video?
A: Modify Video works best with clean, readable footage that gives the model strong motion and structure to track: a steady or smoothly-moving camera, deliberate well-paced action, sharp in-focus and well-lit videos, a single clear subject (or a few well-separated ones) that stays visible and reads distinctly from the background.
Can I change the look/style while keeping the same actor and performance?
A: Yes, you can restyle the look while the original performance, motion and timing carry through.
Does Modify Video preserve lip sync and dialogue timing?
A: Yes, when Face tracking is on.
What video file formats can I use with Modify Video?
A: .mp4, .mov, .wmv
What are the limitations of Ray3.2 Modify Video?
A: Modify Video can struggle with footage which has heavy camera shake, whip-fast or chaotic action, blurry/dark/low-res footage, crowded overlapping subjects, and heavily occluded faces or bodies and smearing, which can cause lost performance fidelity.
Adherence
What does adjusting Motion and Structure settings affect my Modify Video output?
Motion is the movement in your footage: body movement, choreography, and the camera's own dynamics (pans, handheld shake, dolly moves). Modify Video is built to preserve this. You shoot a performance once, and the timing and movement carry through no matter how radically you restyle the look.
Structure is the underlying spatial layout: the shapes, forms, edges, and composition of each frame. It's the scaffolding the new generation is built on top of. Strong structure adherence means the output tracks the original's forms tightly; loose adherence lets the model reinvent those forms.
Characters:
Poses - Up to eight skeletal tracks’ positions, gestures, and postures held across the clip.
Blocking - The loose spatial locations and gross movement of characters, where the character sits in the scene's space: framing and relationship to the surrounding environment, without strong adherence to the source pose
Adherence Controls (Scale 1 to 9)
Auto Adherence is enabled by default. Manual adjustments allow you to fine tune preservation levels:
Motion (1 to 9): Controls movement preservation. Increase Motion when gesture, action, or camera passes must stay locked to the original.
Structure (1 to 9): Controls spatial layout. Increase Structure when object geometry, scene framing, and layout must remain stable.
How does Multi-Keyframe support improve my Modify video workflow?
A: You can now set up to 64 individual keyframes within a single clip. This provides precise control over narrative timing and storybeats, ensuring the model executes your specific vision frame by frame.
Reframe
How do the new Reframing tools work?
A: Reframe expands your videos to new aspect ratios by filling in the frame with scene-aware context. The new Reframe now allows custom positioning within the frame.
What video length can I reframe?
A: You can use Reframe for videos up to 12 seconds long.
Image2video and text2video
What aspect ratios are available?
A: 9:16; 3:4; 1:1; 4:3; 16:9; 21:9
How long are the videos generated with 3.2?
A: 5 seconds and 10 second native generations
How many keyframes can I use for Image2video with Ray3.2?
A: You can add up to 64 keyframes for image2video.
Is the Ray3.2 available via API?
A: Yes, the Ray3.2 model is available as an API. This allows developers, agencies, and studios to integrate the full control surface including multi-keyframe, performance tracking, and EXR output directly into their own tools and workflows.