How a VW Designer Built a 48‑Second Sci‑Fi Trailer with Midjourney, Runway, and CapCut

Background

In early 2024 a 48‑second sci‑fi trailer titled Genesis went viral on social media. The trailer was produced by Nicolas Neubert, a senior product designer at Volkswagen, who used generative AI tools to create every visual, audio, and video element.

Tools Used

Midjourney – image generation (316 prompts, 128 upscaled images)

Runway – video generation (310 video clips created, 44 used in the final cut)

CapCut – final video editing and beat‑synchronisation

Music sourced from Pixabay

Total production time was about 7 hours and the direct cost was roughly $125 (Runway $95, Midjourney $30).

Production Process

1. Choose the soundtrack – The designer selected a piece from Pixabay and used it as the rhythmic backbone for the trailer.

2. Storyboarding – The narrative was divided into three tension‑building phases: setting the scene, introducing a threat, and a climactic call‑to‑action. Small visual motifs (e.g., glowing amulets on child characters) were added to hint at deeper story layers.

3. Prompt design for Midjourney – A reusable template was created to generate consistent visual styles. Example prompt (shown in the original article) is reproduced below:

___________ , Star Wars war, fine crowd scene, rustic naturalism, cyan and yellow, frost‑punk, indoor scene, cinestill 50d --ar 21:9 —original style

The designer filled the blank with scene‑specific keywords and leveraged Midjourney’s Strong Variations feature to keep the colour palette while producing diverse shots (e.g., swapping a female warrior for a civilian or hacker).

4. Video generation in Runway – Each Midjourney image was fed into Runway to create animated clips. The creator preferred the “Upscaled” mode for most shots, switching to “Standard” quality for facial close‑ups to avoid artifacts.

5. Editing in CapCut – The final step involved importing all clips into CapCut, aligning them to the music’s beat, and stitching them together on the timeline to form a coherent story.

Key Takeaways

The project demonstrates that generative AI can dramatically accelerate creative workflows, but human direction remains essential. Prompt engineering, story planning, and iterative refinement are still required to achieve high‑quality results, as fully autonomous AI output often falls short.

For readers interested in replicating the workflow, the author also linked to deeper tutorials on Midjourney and Runway, and noted that Runway’s Gen‑2 model is now freely accessible (4‑second clips cost 5 credits, $0.01 per credit).