Inspiration

This project began with a simple question, sparked by the profound impression left by Netflix's : "What if we could explore the concepts of the multiverse and parallel universes through a story?" This short film follows a man who comes to believe in the multiverse. It portrays his struggle to find his "real" self among countless dimensional counterparts and the confusion that leads him to seek help at a clinic.

What it does

Our goal was clear: to provide a short, impactful experience that audiences could engage with without needing complex background knowledge. The film aims to concisely deliver the philosophical question, "Who Am I?"

How we built it

This film was produced based on a 100% AI workflow. First, we generated key images and visual concepts using Midjourney, Google 2.5 flash (Nanobanana), and Kling AI. Following that, we used Kling AI as our primary tool to create all the video clips. For the character dialogue, lip-syncing was implemented using Kling AI's avatar feature, and the voices were generated with ElevenLabs. The background sounds were also created using ElevenLabs' Sound Effect templates or were custom-made. Finally, all assets were imported into After Effects for post-production editing and compositing.

Challenges we ran into

The production process was not without its hurdles. We faced two significant difficulties, especially when using generative AI models. First, there were limitations in creating a wide variety of cuts. Second, maintaining the consistency of the same character across multiple scenes proved to be our biggest technical challenge.

Accomplishments that we're proud of

Despite these challenges, we consider it a small accomplishment that, after much trial and error, we achieved relatively smooth transitions and scene connections. We strove to deliver the best possible narrative within the available technological constraints.

What we learned

This project was a valuable opportunity to witness firsthand both the clear limitations and the emerging possibilities of current AI video generation technology. We clearly recognized shortcomings, such as the character consistency issue, and it has prompted us to think about new workflows to compensate for them.

What's next for Who Am I

Our next goal is to build upon what we've learned and address the technical shortcomings. We plan to produce a more advanced short film, specifically by improving character consistency and attempting more dynamic cut editing.

Built With

  • aftereffects
  • elvenlabs
  • english
  • klingai
  • mediaencorder
  • midjourney
  • nanobanana
Share this project:

Updates

Submission history