How Google used new AI techniques to enhance 90%+ of The Wizard of Oz, released in 1939, to show on the Las Vegas Sphere's giant screen from August 28, 2025
Google invented new techniques to enhance resolution and generated new character performances to bring the 1939 film to the giant screen
Context & Ripple Effects
AI video enhancement had already been moving from restoration-style upscaling and colorization toward broader image transformation, as shown by earlier AI tools for improving old-film footage. Google’s work applies that trajectory to a premium, venue-specific presentation rather than a general-purpose remaster.
It also arrives as generative filmmaking tools improve while still demanding painstaking work to maintain visual continuity, a constraint highlighted in a test of AI-assisted film production. That makes a feature-length legacy title a meaningful production-scale use case.
First-order effects
- The Las Vegas Sphere receives a version of the film tailored to its unusually large display, with AI-enhanced resolution across more than 90% of the footage and newly generated character performances.
- Google demonstrates a production workflow that goes beyond cleanup or upscaling: it can alter and extend elements of existing screen material for a specific exhibition format.
Second-order effects
- Studios, archives, and exhibitors evaluating large-format or immersive revivals gain a concrete reference point for using generative tools to adapt catalog titles whose original footage was not made for modern displays.
- The project raises the production bar for AI restoration vendors: customers may increasingly expect both technical enhancement and controlled scene-level generation, not simply sharper images.
Third-order effects
- If this model is repeated, the boundary between restoration and adaptation will blur, making provenance, creative approval, and audience disclosure more central to how legacy works are repackaged.
- Premium venues could become an early commercialization channel for generative post-production, where bespoke versions can justify the labor still needed for consistency and quality control.
The trend: Generative AI is moving from creator experiments and restoration tools into commercially packaged, format-specific adaptations of established film libraries.