The tribeca festival 2026 showcased a significant shift in artificial intelligence filmmaking with the premiere of dear upstairs neighbors, a feature directed by simon horsman. Unlike previous attempts that relied on generic generative models, this project was built using custom versions of google deepmind’s veo and openai‘s imagen. The film demonstrates that high-quality production is now possible when studios train specific models on their own historical footage rather than feeding prompts into vanilla systems. This approach marks a departure from the short-form video slop that has dominated recent industry reports and highlights a move toward professional-grade output.
The significance of this development lies in its potential to stabilise the relationship between hollywood and silicon valley technology providers. By creating bespoke models, production houses can ensure consistency and control over their creative assets, avoiding the unreliable nature of off-the-shelf tools. This strategy addresses the growing concern that major partnerships are evaporating because generic models cannot meet the rigorous demands of feature-length storytelling. If this method proves successful, it suggests that the future of cinema depends on deep integration and proprietary training rather than broad, one-size-fits-all applications.
* custom model training allows for consistent visual quality across long-form productions.
* studios are moving away from generic generative tools to protect their creative integrity.
* the tribeca premiere signals a new era of reliable ai integration in hollywood.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
