Inspiration

In tomorrow's ateliers, fashion designers abandon fabric for a radical new medium; consciousness itself. Agents unveils a speculative future of brain-machine interfaces where neural technology becomes wearable art. A cast of cyborgs showcase 'neural veils' merging the organic mind seamlessly with synthetic intelligence like a second skin pulsating with computational power. These cognitive cocoons amplify perception and deepen empathy through their biomorphic embrace. Each piece poses a provocation: What if our future augmentations were as beautiful as they were functional? What if the merger of mind and machine resembled not cold hardware but a luminous second skin—intimate, organic, alive? This is fashion forecasting the future of consciousness itself.

How we built it

While current generative AI systems can visualize virtually any textual description, they often produce images with a hyperbolic, kitschy quality that lacks visual sophistication. The Agents collection was generated using our novel Implicit Variation technique (CVPR 2025), which creates visual abstraction through iterative vision-language feedback loops. Rather than direct prompting, we:

  • Generate initial imagery from conceptual metaphors ("neural veils," "cognitive cocoons")
  • Extract emergent visual patterns and semiotic cues from the output
  • Retranslate these visual discoveries into refined language
  • Regenerate with enhanced symbolic emphasis
  • Repeat until meaning crystallizes

This cyclical interplay between vision and language domains allows us to discover visual forms that transcend literal representation—sculpting abstract concepts into tangible aesthetics through generative iteration. The technique's key insight: AI's dual fluency in image and text becomes the creative medium itself, where meaning emerges from the translation between modalities rather than from either domain alone

Challenges we ran into

Creating a speculative fashion film with custom AI-generated imagery proved more resource-intensive than anticipated. Between compute costs for iterative generation, post-production, and the sheer number of variations needed to achieve our aesthetic vision, we exhausted our initial budget, resulting in a shortened music video. We hope to extend this version in the future with additional funding.

Accomplishments that we're proud of

Despite financial constraints, we successfully completed a cohesive video that demonstrates both our technical innovation and artistic vision. The final film showcases the full potential of using AI as a medium for visual abstraction

What we learned

All AI video generation platforms currently lack features for granular control—frame composition, spatial consistency, and precise camera movements remain elusive.

What's next for Agents

The full music video + 3D avatars bringing our cyborg cast to life in virtual reality. We're are also seeking funding to transform this proof-of-concept into physical prototypes in partnership with fashion houses and fabrication studios.

Built With

  • capcut
  • midjourney
  • veo
Share this project:

Updates

Submission history