Devpost Link

Location-specific photo mural where memory persists.

In the physical world, places hold memories. But today’s social interactions are mostly detached from location and time—optimized for close friends or fast-scrolling feeds. We saw an opportunity to create a persistent, location-based, anonymous social layer on top of the real world—one that grows richer the more people participate.

• Instead of swiping, discover anonymous people’s memories by walking around.
• Instead of typing, voice out remixes of other people’s moments.

Sometimes that remix is playful—a time-gap meme or visual joke. Other times, it’s meaningful—like visualizing rising sea levels along a coastline or tracking how a place changes through climate events.

• Discover
  As users walk around, they see photo bubbles anchored to locations, synced from Snap Cloud. Scroll through the mural to explore what others captured there.

• React
  Tap to enlarge, double-tap to like. All interactions are saved to Snap Cloud in real time.

• Remix
  Hit Remix on any photo, voice-prompt our Gemini AI integration, and create a transformed version. The original stays; the remix appears as a new bubble.

• Capture
  Take photos, optionally AI-transform them, and add them to the location’s permanent collection.

Over time, Glimpse becomes more valuable not because of virality, but because it accumulates collective context.

Using Snap Spectacles + Snap Cloud, we turned fleeting interactions into a living, evolving photo mural—a shared memory space where every action permanently alters the world state.

We began by clearly defining the problem, then moved quickly into designing a detailed user flow. This helped us:

• Establish the project scope
• Identify core MVP features
• Separate MVP from nice-to-have enhancements

Once the MVP was solidified, we explored ideas for future expansion and scalability.

Technical stack & approach

• Built in Lens Studio as a mixed-reality application on Snap Spectacles
• Persistent spatial data and interactions saved to Snap Cloud, leveraging PostgreSQL database, real-time websocket, and Deno edge functions.
• Location-aware content and spatial interactions blended with the real world using Snap's Location API.
• Nano Banana integrated for image generation and enhancement during remix workflows

• Amanda

  • Scoping and topic selection
  • Narrowing ideas (sketch, music, photo) into final MVP user steps
  • Presentation framing and video storytelling
  • Overall system architecture and experience cohesion

• Jonathan

  • Integrating location-based photo grouping
  • Distinguishing and ranking most-liked photos

• Sarah

  • Gaining confidence working with Cursor and connecting it to Lens Studio AI
  • Navigating uncertainty around AI tooling due to a non-traditional software background
  • Overcoming skepticism toward AI reliability with the help of teammates and Snap engineers

• Jingle

  • Lack of familiarity with coding and scripting in Lens Studio
  • Understanding syntax and development logic
  • Successfully building required features using vibe coding tools despite limited prior experience

• Team

  • First-time experience with Lens Studio for some members
  • Tooling friction and setup issues on Windows laptops

• Rapidly learning and applying vibe-coding tools (Cursor, Claude Code) and Snap Cloud within Lens Studio
• Designing and implementing the full end-to-end user flow:
  • UI
  • Voice-to-text
  • Databasing
  • Physics-based spatial interactions
• Successfully integrating Nano Banana for user-requested image modification
• Delivering a fully working, functional (and not broken!) prototype

• The importance of fleshing out story, purpose, and user flow early
• How strong upfront alignment enables:
  • Clear feature prioritization
  • Better work distribution
  • Faster end-to-end prototyping
• How to translate prior Unity experience into Lens Studio to build functional mixed-reality features quickly

• Deeper persistence & time layers
  Let users scrub through time to see how a location’s mural evolved over days, months, or years.

• Richer remix tools
  Expand voice-driven remixing into multi-step transformations—visual annotations, generative overlays, and environmental simulations.

• Collective storytelling modes
  Introduce themed prompts (e.g. first memory here, before & after, climate change) to guide meaningful contributions.

• Scalability & performance
  Improve spatial clustering, ranking, and loading for high-density locations.

• Community governance
  Lightweight moderation, reputation signals, and opt-in curation—without breaking anonymity.

Glimpse’s long-term vision is a shared, anonymous memory layer for the physical world—where places don’t just exist, they remember.