Introducing Ray2

January 15, 2025

Next-Generation Video AI

Ray2 is our most advanced video generation model, built on Luma's new multi-modal architecture with 10x the compute power of Ray1. This large-scale video generative model creates photorealistic visuals with natural, coherent motion and has a strong understanding of text instructions.

Key Improvements

Ray2 delivers dramatically enhanced capabilities through advanced architecture and increased compute power. The model produces fast coherent motion, ultra-realistic details, and logical event sequences that increase the success rate of usable generations.

Quality Enhancements: Lifelike textures, smooth camera work, and realistic lighting with physically accurate interactions between objects and characters.

Technical Advances: Higher resolution support, longer video duration, enhanced motion quality with reduced artifacts, and better text prompt understanding for more accurate results.

Production Ready: Streamlined workflow eliminating slow-motion playback issues, faster generation times, and professional-grade results suitable for production use.

Creative Features

Prompting Framework

  • Structured prompting: Main subject → Action → Details → Scene → Style → Camera → Reinforcer
  • Natural language processing with specific, visual descriptions
  • Camera movement integration (crane, tracking shot, dolly, etc.)
  • Style specification with cinematic and lighting controls

Loop & Extension

  • Seamless looping via prompt inclusion or UI toggle
  • Smart extension that maintains motion coherence
  • Quality preservation up to 30-second extensions

Technical Specifications

Output Formats

  • Resolution options: 540p, 720p, 1080p
  • Video length: 5-10 second base clips
  • Extension capability: Up to 30 seconds
  • Upscaling: 4K output available

Platform Availability

  • Web platform: Full feature access via dreammachine.lumalabs.ai
  • iOS app: Mobile-optimized experience with all features
  • API access: For developers and enterprise users

Access Requirements

Ray2 is available to paid Dream Machine subscribers.

Access Ray2 by:

  • Logging into your Dream Machine account
  • Selecting Ray2 from the Description Mode in the Prompt Field
  • Available on both web and iOS platforms

Technical Details

  • Model: Ray2 with multi-modal architecture
  • Training: 10x compute scale of Ray1, trained directly on video data
  • Video Length: 5-10 second base clips, extendable to 30 seconds
  • Resolution: 540p, 720p, 1080p with 4K upscaling available
  • Audio: Supported with prompt-driven generation
  • Features: Keyframes, Extend, Loop, Camera Motion, Style Reference

Support & Feedback

Changelog & Announcements

Stay updated on new features and improvements by re-visiting this changelog or get the latest insights through our news blog.

Tutorials & Learning Hub

Explore our Learning Hub full of in-depth guides and video walkthroughs tailored to supporting creative workflows from experiencing to experiencing to expert!

Feedback & Bug Reports

Share your thoughts and report issues to support@lumalabs.ai. Your input helps us shape the future of Dream Machine.

Enjoy Dream Machine!

Final Words

We're excited to have you as part of the Dream Machine community. Dive into the tools, experiment with workflows, and let your imagination guide you. From creating stunning visuals to animating unique videos, Dream Machine is here to turn your ideas into reality. Let's create something amazing together!