Inspiration

Event hosts take hundreds of photos but never know how the room really felt. Cloud AI can read faces, but it uploads private photos to servers. I wanted a Skill that understands a whole room - emotion, energy, frequency - while keeping every face on the owner's own UGREEN AI NAS.

What it does

EmpathaVision NAS Skill scans event photos and videos stored on UGREEN AI NAS, then returns a Clairvoyant Empathy Report:

  • Empathy Index 0-100, Synchronicity %, Resonance Hz (386 / 432 / 528 / 639)
  • Dominant Affect, Unspoken Narrative, Aura Palette
  • Lighting suggestion in Kelvin, Audio suggestion in Hz/dB, Speaker Empathy Cue
  • 3 auto triggers: Lighting, Audio, Speaker Pause
  • Markdown report + sealed .enc file with AES-256-GCM + SHA-256 seal, verified 100%

Example: A dark immersive room scored Empathy 89/100, 528Hz, Synchronicity 46%. The Skill suggested Amber light + 528Hz sound + 5-second speaker pause to lift synchronicity from 46% to 78% in 2 minutes.

How we built it

Python orchestration with 5 agents: Discover, Perceive, Reason, Prescribe, Report and Seal.

  • Discover: UGREEN-files list authorized media
  • Perceive: local vision analysis for entities, gaze, lighting, mood
  • Reason: aggregate into collective metrics
  • Prescribe: map metrics to triggers
  • Report and Seal: write Markdown, encrypt JSON with AES-256-GCM, verify SHA-256 No faces leave the NAS. No external API in the core loop.

Challenges we ran into

Dark rooms are hard - low light hides faces. We solved it with spatial mood + lighting Kelvin analysis, not just face detection. Sealing was hard too - we had to make round-trip verification 100% or delete the file.

Accomplishments that we're proud of

  • 7 sealed sessions, Platinum seal, SHA-256 match 100%
  • Empathy Index up to 89/100 on real event photos
  • Full local privacy: no cloud upload

What we learned

Empathy is not carrying other people's sadness. It is understanding without burnout. Technology should protect privacy while helping people feel seen.

What's next for EmpathaVision

Test on real UGREEN AI NAS iDX6011 Pro, add video frame sampling, and support Vietnamese + English reports.

Built With

Share this project:

Updates

posted an update —

Day 1 Update - EmpathaVision AI is LIVE on UGREEN Hackathon!

From Cho Lach, Ben Tre to the world!

Today I submitted EmpathaVision AI - 528Hz Clairvoyant Empathy OS to UGREEN Hackathon 2026 on AI NAS - Track: Explore Your Imagination.

What I built today:

  • Full Proposal PDF - 10 sections, under 500 words, with real test data
  • SKILL.md for UGREEN NAS - empathavision-nas-skill - local, private, sealed
  • 5-Agent Flow: Discover -> Perceive -> Reason -> Prescribe -> Report and Seal

Real test result:

Dark exhibition, 4 entities detected.

  • Empathy: 89/100
  • Resonance: 528Hz
  • Synchronicity: 46% -> 78% in 2 minutes
  • Lighting: Blackout -> Deep Violet #3B1E54
  • Audio: Binaural 4.5Hz Theta + Cello on 528Hz drone
  • MC Cue: "Darkness is not forgetting, it is where life begins to define itself."

Privacy first:

No faces leave the NAS. Reports are sealed with AES-256-GCM + SHA-256, Platinum v5.2, 100% tag match. 7 sessions sealed, all OFFLINE & ENCRYPTED.

What I learned:

Empathy is not carrying sadness. It is understanding without burning out.

Next: Test on 【entity-UGREEN¦canonical_name=UGREEN】 iDX6011 Pro, add video sampling, Vietnamese + English support.

Thank you for following my journey from a small garden in Ben Tre! Namaste! - Thai Thanh Nguyen, NucleusNexus Lab

Submission history