Use your phone to create 3D renders of objects for video. Epic results. Capture in lifelike 3D. Unmatched photorealism, reflections, and details. The future of VFX is now, for everyone!
This is the 7th launch from Luma AI. View more
Luma Ray 3
The world’s first reasoning video model, and the first to generate studio-grade HDR. Now with an all-new Draft Mode for rapid iteration in creative workflows, and state of the art physics and consistency.









Free Options
Launch Team






FLORA ©
Melies
Hey Hunters 👋
I am excited to hunt Ray3 — the world’s first reasoning video model, and the first to generate studio-grade HDR.
🚀 What’s new in Ray3:
✅ Reasoning engine that thinks in visuals + language
✅ All-new Draft Mode for rapid iteration
✅ State-of-the-art physics and scene consistency
✅ Visual annotations to direct motion, blocking & camera
✅ Native 10/12/16-bit HDR for stunning color depth
✅ Production-ready motion, crowds, lighting, caustics, motion blur, and more
With reasoning, Ray3 understands nuanced directions, judges its outputs, and creates complex multi-step motion faster than ever — giving you precise control and reliably better results.
Start creating today
https://lumalabs.ai/ray
https://lumalabs.ai/dream-machine
Impressed with SOTA scene consistency and comprehensibility of the physical world of Luma Ray 3! Can't wait to try it out!
BigMotion AI Slideshow
Hey Luma AI team,
Big congrats on the 3.0 launch - excited to keep using it in our projects via the API :D
We’re also live on Product Hunt today with BigMotion.ai (organic ads), where we integrated Luma AI for creating UGC Hooks
Scrumball
The reasoning engine approach is interesting - most video AI tools generate based on prompts but can't really understand or evaluate their own output quality.
How does the Draft Mode actually work for iteration? Can you make specific changes to scenes without regenerating the entire video, or is it more about quickly testing different prompt variations?
The HDR claims sound impressive but how does it perform with complex lighting scenarios? Most AI video still struggles with realistic shadows and reflections in multi-object scenes