Luma Ray 3.2: photorealistic AI video with frame-level control

Luma Ray 3.2 is Luma Labs’ latest AI video model, built for precision and photorealism. It turns text prompts or a reference image into cinematic clips up to 20 seconds long at 1080p, with frame-level multi-keyframe control for directing motion and pacing. A native HDR pipeline with 16-bit EXR output keeps lighting, reflections, and materials physically accurate, making Ray 3.2 a production-grade model for generating believable, high-fidelity video.

Ray 3.2 pairs photorealistic generation with real directorial control. Place keyframes across a clip to choreograph motion and timing exactly, generate from text or an image, and render up to 1080p for clips as long as 20 seconds. Its native HDR pipeline produces 16-bit EXR-ready output for color grading and compositing, while expressive performance tracking preserves skeletal posture, gestures, and facial nuance for multiple subjects at once — so complex, human scenes stay convincing frame by frame.

Picsart integrates Luma Ray 3.2 directly into its AI Playground and AI Video Generator, so you can generate video from text or an image without touching the model directly. It works alongside tools like the AI Voice Generator and AI Video Editor, helping you take a generated clip all the way to a finished, platform-ready project.

Luma Ray 3.2 gives creators photorealistic quality and the control to direct it. They choose it for frame-level keyframe direction, native HDR output, and cinematic 1080p clips up to 20 seconds that hold up in real production. Integrated into Picsart’s AI Video Generator, it makes high-fidelity, directable video generation accessible without complex setups or specialized equipment.