Generating...

Clueless meets Black Mirror. "Generating..." is a comedic short film with a dark sci-fi twist. Two ambitious women compete in a cutthroat corporate world where generative reality is part of life.

Comment

Inspiration

I was inspired by dark satirical sci-fi such as Black Mirror. I wanted to create a seemingly happy story/character that is hiding a dark reality. The name for the Chroma-Gen "generative reality assistant" was inspired directly by The Chroma Awards itself. The character Jen was inspired by Alicia Silverstone's character from Clueless. The dialogue and story arc is a bit of a nod to the campy, corporate rivalry tropes from Chinese micro-dramas.

How we built it

This short film was generated and edited by one person over the course of 8 days, with the assistance of a variety of AI tools including VEO 3.1, Kling AI, OpenArt, NanoBanana, and ElevenLabs. The film was edited in Davinci Resolve. Licensed Music Provided by Epidemic Sound ("Expanding" by Elliptik).

  • VEO 3.1 (I2V, Lip Sync, Voices of Jen/Nina/Boss, Start-to-End Frames)
  • Kling 2.1/2.5 (Start-to-End Frames, I2V)
  • NanoBanana (Character Design, Environments, Image Gen, Image Editing)
  • OpenArt (Image Gen, Image Editing)
  • ElevenLabs (Voice of Chroma-Gen and TV Announcers)
  • Sora 2 (T2V, Voice of Voice Actor in End Credits Scene)
  • Flux Kontext Pro (Image Gen)
  • Seedream 4.0 (Image Gen)
  • Seedance 1.0 (Start-to-End Frames)

Challenges we ran into

Character consistency/drift was the biggest challenge. No matter how hard I tried to keep my characters consistent, the video generator would ultimately morph their likeness to someone slightly different each time. It felt like a losing battle, so instead I focused on getting good performances out of the AI characters, even if the character consistency was off a bit sometimes.

Accomplishments that we're proud of

I wanted to focus on implementing high quality AI lip sync, good AI voice performances, realistic facial expressions, proper camera positioning, and good spatial coherence of characters moving within environments. I was really happy with the level of realism that was achieved from all of those elements working together.

What we learned

Continued to improve on my Start-to-End Frame techniques. I also learned a lot about camera positioning with AI tools, and how to connect shots together to form a cohesive, linear scene that makes sense within time and space. I learned that Start-to-End Frame shots are quite good at doing orbit transitions, which can be an effective way to establish spatial cohesion within a scene, as well as transition between scenes creatively.

What's next for Generating

Wonder Film Festival is next!

Built With

  • ai
Share this project:

Updates