Lumiere Space-Time Video Generation Quiz

Answers and explanations appear as you take the quiz.

1. What is the defining architectural feature of Lumiere?
   • A Space-Time U-Net that processes the whole clip at once
   • A keyframe-only generator with no interpolation
   • A purely text-based language model
   • A single-frame image generator
2. How does Lumiere’s temporal generation strategy differ from sparse-keyframe pipelines?
   • It only outputs static images
   • It generates the full temporal duration in one model pass rather than filling between distant keyframes
   • It generates sparse keyframes and performs no temporal processing
   • It requires a 3D scan as input
3. What challenge does Lumiere’s full-duration design aim to address?
   • Audio quality
   • Image file compression
   • Global temporal consistency across generated motion
   • Spelling in captions
4. In what dimensions does the Space-Time U-Net downsample?
   • Only color channels
   • Only the time axis
   • Neither space nor time
   • Both space and time
5. Which creative effect, animating only part of a still image, does Lumiere support?
   • Cinemagraphs
   • Podcasts
   • 3D printing
   • OCR text extraction
6. Which organization introduced Lumiere?
   • A university student club
   • Google Research
   • An anonymous open-source group with no affiliation
   • A hardware chip maker
7. What did the Lumiere paper use to initialize its video diffusion model?
   • A speech-recognition transformer
   • A neural radiance field trained only on 3D scans
   • A pretrained text-to-image diffusion model
   • A rule-based animation engine
8. What kind of clip does Lumiere directly generate before spatial upsampling?
   • Only a single still image
   • A sparse set of unrelated keyframes
   • A text transcript with no frames
   • A low-resolution, full-frame-rate video