Inspiration
This project started from a simple “what if”: a serious spec-ops time-travel mission that secretly runs like a reality show. Cameras in the trees, producers watching every move, and soldiers who only realise they’re being filmed when it’s almost too late. We wanted the tension of dinosaurs and tactical ops, plus the horror of discovering you were entertainment all along.
What it does
The story follows an elite team dropped into the Cretaceous under a strict mission protocol.
Objectives seem clear—until the squad notices hidden optics in the jungle, strange angles in the trees, and patterns that feel staged.
They realise:
- the environment is filled with disguised cameras and drones,
- every “choice” is being recorded and escalated,
- someone expects them not to come back.
How we built it
We outlined the core beats: drop, first anomaly, discovery of the cameras, protocol turning hostile, and the decision to break out instead of play along.
For visuals we used AI image tools to design:
- the squad (gear, silhouettes, face variety),
- prehistoric terrain and predators,
- camouflaged devices and structures hidden in foliage.
Challenges we ran into
- Keeping the team, their gear and weapons consistent across different AI models.
- Making the Cretaceous environment feel dense and alive without losing clarity in 9:16.
- Showing the hidden “show” layer visually (cameras, lenses, strange angles) without over-explaining it.
- Balancing creature action, gunfire and comms so it stays readable on a phone screen.
Accomplishments that we're proud of
- We built a clear hook and twist: the moment the squad notices they’re being watched and decides to stop following the script.
- We defined a look where military visuals, dinosaurs and covert tech feel like one world instead of separate elements.
- We created a reusable pipeline for vertical action with creatures, VFX-style moments and layered sound.
What we learned
We learned how important it is to signal the “hidden camera” idea early using composition and props, not just dialogue.
We saw that for vertical action, clean silhouettes and simple geography matter more than complexity in each frame.
We also learned how to guide AI tools toward consistent squad members and recurring predators across multiple sequences.
What's next for the project
Next, we plan to expand this into a sequence of short “episodes” where each objective is clearly staged for the unseen audience—and the team starts actively sabotaging the show.
Built With
- aleph
- banana
- capcut
- dreamina
- elevenlabs-sfx
- elevenlabs-tts
- fashion
- flux
- fluxk
- hailuo-2
- hidream
- higgsfield
- ideogram
- imagen
- kling
- ltx-studio
- luma
- minimax-tts
- omnihuman
- openai
- pixverse
- pixverse-lipsync
- q1-ref
- qwen
- runway
- sdream
- sedit
- seedance
- sora2
- thinksound
- topaz-upscale
- veo
- wan

Log in or sign up for Devpost to join the conversation.