Dream Lens

An AI-driven VR documentary that transforms recorded observations into personalized stories.

Concept Design

Dream Lens was created during an AI × XR hackathon by a team with experience in VR, AR, and game development. It is a playable, AI-driven VR documentary that turns player-recorded observations into personalized documentary-style narratives.

The experience moves between two contrasting environments: a warm African savannah and a quiet Arctic landscape. Rather than simply documenting these worlds, players collect what catches their attention and use it to shape a story of their own.

Gameplay Flow

Enter the scene — step into the savannah or Arctic environment.
Record the world — frame and capture observations with a virtual camera.
Detect tagged objects — identify the objects collected during the recording.
Generate the story — use Gemini to turn those observations into a short, personalized narrative and mood.
Voice and feedback — hear the story through Meta SDK text-to-speech, supported by ambience and sound effects.

AI-Driven Mood Filter System

Dual-Camera Interaction

Players can switch between two cameras. The reality camera preserves the original scene, while the imaginative camera transforms it through AI-driven mood filters.

Mood-Driven Filtering

Gemini's generated mood is mapped to four states in Unity — Happy, Sad, Dream, and Satanic — dynamically selecting the visual treatment for the imaginative camera.

The result is a deliberate contrast between observation and interpretation: one lens records what is present, while the other reveals how the generated story feels.

My Role

Scene Assembly

I built and dressed the African savannah and Arctic scenes in Unity, focusing on composition, asset placement, lighting, and atmosphere.

AI Narration & Audio Pipeline

I connected Gemini story generation with Meta SDK text-to-speech to create a runtime narration pipeline in Unity. I also handled voice-over, ambience, and sound effects to support the documentary-style presentation.

Experience Highlights

Active Meaning-Making: Players move from passive observers to active creators of narrative through their attention and selection.
Dual Environments: The warm, vibrant savannah contrasts with the quiet, introspective Arctic landscape.
Immersive Interpretation: A reimagined documentary format where the world is not just recorded, but emotionally transformed.

Award

Hosts’ Favourite Award — 2026 Cambridge Reality Hackathon