Linus Konetschnig
ABOUT
Launch film for Lucent Chat, an agentic video generation platform later acquired by fal.ai. I made it in 2025, before Veo 3.1 and before Seedance took the lead in AI video, so every shot had to be pulled out of the models available at the time. Some of them took eight hours of JSON prompt engineering to land. The film follows a tarsier from a cave wall to the Colosseum, a concert hall and deep space: cinematic storytelling built with tools that were never designed for it.
SELECTED FRAMES — 4 OF 8
PROCESS
Before Seedance changed the pace of AI video generation, Veo was the flagship model. I created this piece before Veo 3.1 launched, in a very different era of the process: hours spent engineering JSON prompts with GPT-4.0, followed by thousands of generated frames and around $3,000 in generation credits alone to reach the quality I was aiming for. Today’s models have moved the goalposts. This clip remains a time capsule of the experimentation, persistence and craft it took to make ambitious AI video at the time.
Credits
More Film Work
Los Resucitados
Esteban Diaz Baena
Altered State
Ethereal Moon
While We Are Here
Erik Luo