Ash Koosha Debuts All-AI Film at Tribeca Festival

Dreams of Violets, a 75-minute all-AI feature by brothers Ash and Pooya Koosha, premiered at the 2026 Tribeca Festival on June 10, per Variety and Deadline. The Hollywood Reporter reports the film used Kling AI for video, Anthropic's Claude for language editing, Google's Gemini and Nanobanana for research and imagery, and Fountain 0's own tools - Koosha described the process as "nearly impossible." CBC and other outlets report a reported $2,000 production budget. The film fictionally dramatizes the January 2026 Iranian civilian crackdown. Tribeca co-founder Jane Rosenthal defended the festival's decision, per Variety and CBC, while the premiere has generated widespread online backlash.
What happened
Dreams of Violets, a full-length, live-action feature created without cameras or actors, premiered at the 2026 Tribeca Festival on June 10, per Deadline and Variety. The Hollywood Reporter reports every image and character in the film was generated using a mix of AI tools - specifically Kling AI for video generation, Anthropic's Claude for language-related editing, Google's Gemini and Nanobanana for research and imagery, and Fountain 0's own technology for blocking and frame accuracy. Deadline and Variety describe the film as a fictional dramatization of the January 2026 crackdown on Iranian civilians, which international organizations have characterized as involving thousands of deaths. CBC and other outlets report the brothers produced the movie for about $2,000. The Hollywood Reporter records Ash Koosha saying, "I slept three hours a night, I was having major headaches, it was nearly impossible." Variety and CBC quote Tribeca co-founder Jane Rosenthal defending the festival choice: "If somebody wrote a song about it, you wouldn't say anything, if somebody wrote a poem about it, you wouldn't say anything, if somebody wanted to dance about it, you wouldn't say anything." Multiple outlets record social-media backlash to the film's AI origin.
Technical details
Public reporting identifies this project as assembled from AI-generated visual and narrative assets rather than traditional on-set production. The Hollywood Reporter lists the specific pipeline: Kling AI (video), Claude (language/editing), Gemini and Nanobanana (research and imagery), and Fountain 0's own tools (blocking and frame accuracy). Projects that stitch together generated frames, synthesized audio, and LLM-driven scripts require heavy human orchestration for continuity, pacing, and artifact mitigation, which helps explain the long iterative loop Koosha describes in interviews.
Context and significance
Festival acceptance of an all-AI, live-action feature places generative-video output into mainstream exhibition and intensifies debates around ethics, provenance, and consent. Reporting across The Hollywood Reporter, Variety, Deadline, and CBC frames the premiere as historically notable - several outlets call it the first full-length, live-action AI-generated film accepted by a major festival. The reported $2,000 production budget highlights how generative pipelines can collapse traditional production costs, raising questions about labor displacement, crediting, and the economics of visual effects and independent film financing.
What to watch
For practitioners: dataset and rights scrutiny - whether source material used to train or drive the film's generative components has been disclosed. Platform and vendor policy response: statements from Anthropic and Google about permitted uses and model-card disclosures. Festival and distribution standards: whether programmers, distributors, and guilds adapt submission rules or crediting norms for synthetic actors and assets. Technical audits: independent analyses that test for artifacts, hallucinated details, or misuse of protected likenesses.
Scoring Rationale
The Tribeca premiere of an all-AI live-action feature is a genuine first for a major festival, with multi-source reporting and direct implications for generative-video production economics, provenance, and platform policy. Well-sourced and notable for AI/media practitioners; a modest pullback from 7.2 reflects that the film's cultural moment is clearer than its technical verification.
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


