# Google DeepMind showcases generative AI at Tribeca with Dear Upstairs Neighbors > Source: > Published: 2026-06-13 11:51:29.632627+00:00 # Google DeepMind showcases generative AI at Tribeca with Dear Upstairs Neighbors The Verge reports that Google DeepMind's animated short "Dear Upstairs Neighbors" screened at the Tribeca Film Festival and is presented as a case study in using generative AI for creative filmmaking. The Verge notes concept art from the film was used to train custom builds of Google's Veo and Imagen models. Coverage contrasts these curated, hybrid projects with the low-quality, 'prompt-to-video' outputs that many current video models still generate, and says several Tribeca entries demonstrated how human filmmakers can leverage generative tools for concepting, imagery, and montage rather than producing feature-length films purely from vanilla prompts. The Verge also describes some festival projects as uneven, with a few examples feeling disjointed or lacking polish, illustrating current limits alongside creative possibilities. ### What happened The Verge reports that Google DeepMind's animated short "Dear Upstairs Neighbors" is screening at the Tribeca Film Festival 2026 as a case study in creative uses of generative AI. The film - directed by Pixar alum Connie He and first presented at Sundance in January 2026 - used concept art to fine-tune custom builds of Google's Veo and Imagen models, then employed video-to-video workflows to let animators convey precise motion and comedic timing. The Verge covered multiple AI-powered films at Tribeca, contrasting curated, human-led projects against lower-quality, prompt-driven video outputs seen elsewhere online. ### Technical workflow According to Google's own production writeup (blog.google, January 2026), the team built tools allowing artists to fine-tune Veo and Imagen on small sets of bespoke concept-art images, teaching the models character design and visual style. A video-to-video pipeline then transformed rough 3D or 2D animation (created in Maya or TVPaint) into stylized generative output, preserving motion intent while applying the film's expressionistic look. Localized refinement tools let artists iterate on specific video regions without regenerating full shots. Final frames were upscaled to 4K via Veo's upscaling capability. ### Broader Tribeca context "Dear Upstairs Neighbors" is one of several AI-infused films screening at Tribeca 2026. A separate entry, "Dreams of Violets" - directed by brothers Ash and Pooya Koosha and reported by Deadline (May 2026) - is a feature-length film in which every image and person was generated using AI video tools, built around journalistic accounts of events in Iran. The presence of both a high-craft, hybrid production and a fully AI-generated feature at the same festival illustrates the breadth of current generative-video adoption in creative contexts. ### So what for practitioners The Verge's framing underscores a distinction that practitioners already observe: generative models deliver cinematic-quality results when combined with artist-led fine-tuning, bespoke datasets, and traditional post-production craft - not when used as end-to-end content factories via off-the-shelf prompts. For teams integrating Veo, Imagen, or similar models into production, the "Dear Upstairs Neighbors" workflow offers a documented example of how custom model fine-tuning and video-to-video pipelines can extend, rather than replace, animator control. ## Scoring Rationale The story documents a well-executed generative-video production workflow at a visible festival, with primary-source technical detail from Google's own production writeup. Relevant to practitioners integrating Veo or similar models, but represents incremental adoption of established fine-tuning and video-to-video techniques rather than a novel breakthrough. Original n8n score of 6.8 pulled to 6.0 to reflect single Verge source at publication and a showcase story rather than a technical advance. Practice interview problems based on real data 1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with. [Try 250 free problems](/problems)