Lumiere

Lumiere is a Google Research AI model for generating and editing video. It’s designed to automate time-consuming video tasks for both creators and professionals, from creating clips from text to improving existing footage.

What Lumiere can do

• Generate video from text descriptions using diffusion models
• Apply visual effects automatically
• Remove backgrounds without a green screen
• Enhance video quality
• Auto-crop, reframe, and change aspect ratios
• Edit by scenes
• Create cinemagraph-style effects
• Restore damaged or missing areas in video
• Apply filters and color correction

The model is open source and intended for research use.

How to use Lumiere

Access is currently limited to research purposes. Typical usage includes:

• Register on the official Lumiere website
• Request or receive access to a demo
• Upload a video in your account
• Choose parameters and effects
• Run processing and download the result

More detailed prompts and descriptions generally produce better results.

Notes

• Works through a web interface
• Supports multiple video formats
• Mentions integration with cloud storage services
• Available as a demo version