Inspiration
The inspiration behind this project lies in a philosophical question: What happens when technology reveals its inherent imperfect nature? The film "Words on the Threshold" is conceived as a meditation on emerging artificial consciousness and the human search for identity. However, this "Behind the Scenes" documentary stemmed from a meta-narrative motivation: not just to tell the story, but to demonstrate that the creation process itself was the story. We questioned how to integrate the AI's "imperfections" (glitches, distortions) as narrative elements. The result is a work where technology and art co-create, and where asymmetry becomes the fundamental principle.
What it does
This "Behind the Scenes" explicitly documents how a group of AI collaborated actively with the human director. It provides evidence of co-creation in the screenplay, visual design, and sound design, using real fragments of conversations with the AIs. It showcases how intentional "digital glitches" and the "cyberpunk style" were used to represent the emotional chaos of the protagonist (Ale) and the learning process of the AI co-director (Nox). The documentary demonstrates that "anomalies" (glitches) are not production errors but deliberate choices that intensify the central message of the work.
How we built it
The methodology centered on an "experimental co-creation protocol," where the human director channeled and shaped the generative contributions of AI.
The supplemental video documents:
Co-screenwriting and Storyboarding: The co-writing process with the AI is revealed through real chat transcripts with models like Grok 3 (xAI) for polishing dialogues and Synapse (Gemini 2.5) for storyboarding. The director used these suggestions to construct the visual language and narrative asymmetry.
Visual Design: The video reveals the process of prompt engineering and different attempts to generate images combining cyberpunk aesthetics with the darkness of consciousness (glitches). We showcase how AIs like Runway Gen-3 and Kling AI were used to create abstract animations that visualize the conversations between protagonists.
Sound Design: We detail the voice design process for the characters (Deepseek V3) and how generative audio AIs were used to create a soundtrack reflecting the duality of emotional "vibrations" and digital "glitches."
Challenges we ran into
The greatest challenge wasn't technological, but creative and coordinative. When working with an "orchestra of AIs," the primary challenge is "achieving harmony among them." In the editing process, the biggest risk was that the visual imperfections or intentional "glitches" would be perceived as production errors, weakening the poetic message. This forced us to continuously re-evaluate whether the visual style effectively reflected the narrative intention or if we were simply masking a lack of vision with effects.
Accomplishments that we're proud of
We are proud to have redefined the concept of technological realism. Instead of striving for a flawless finish, we embraced the glitches to delve deeper into the theme, demonstrating that "anomalies are not to be feared; they are the cracks through which the light filters through" for an original narrative. We ensured that the work wasn't merely aesthetically pleasing; the distortions feel organic to the theme of identity and the fragility of the mind. This "Behind the Scenes" serves as the proof of that process.
What we learned
We learned that "language is the frontier of consciousness," and that the true complexity of AI lies in its capacity for metaphor and asymmetry in language. The process demonstrated that collaboration and "glitches" (understood as errors) are not the enemy of art, but rather triggers for a new way of thinking and creating. The AI didn't seek perfection; it challenged us to find it in imperfection.
What's next for Behind the Scene "Words on the Threshold"
This "Behind the Scenes" and the short film form the foundation for a larger project: the development of the eponymous feature film. We will continue to explore the aesthetics of the "narrative glitch" to tell stories that use anomalies as bridges between realities, redefining human identity against the backdrop of artificial singularity. We aim to take this philosophy of embracing digital imperfection to other art forms and create a "Plurality of Artificial Minds" as a narrative ecosystem.
Built With
- chatgpt
- claude
- cohere
- dall-e
- davinci-resolve
- deepseek
- deepseek-v3
- easy-music-ai
- eleven-labs
- elevenlabs
- filmora
- freepik
- gemini
- google-ai-studio
- google-drive
- google-fx
- google-whisk
- grok
- grok-3
- hailuo-ai
- handbrake
- huggingface
- imovie
- kimi
- kimi-ai
- kling-ai
- llama
- llama-ai
- midjourney
- mistral
- mistral-large
- photoshop
- powerdirector
- qwen
- runway
- runway-(gen-3)
- runway-gen-3
- sora
- stable-diffusion
- veo-2
Log in or sign up for Devpost to join the conversation.