PHILLY_IN_FLUX_AUTOMATION_VISION

Core Goal

Build a self-expanding documentary system capable of continuously archiving Philadelphia through walking and photography.

The objective is not simply to create photographs.

The objective is to create a living archive of Philadelphia that grows automatically over time.


Long-Term Vision

Walk
→ Photograph
→ Import
→ AI-assisted processing
→ Project generation
→ Publication
→ Archive expansion

The photographer focuses on:

  • walking
  • seeing
  • exploring
  • photographing
  • final editing

The system handles:

  • organization
  • mapping
  • statistics
  • project generation
  • archive maintenance

Philly In Flux

Philadelphia In Flux is a city-scale photographic survey of Philadelphia.

Every completed walk contributes to a larger archive:

  • Market Street
  • Broad Street
  • Frankford Ave
  • Germantown Ave
  • Washington Ave
  • Ridge Ave
  • Future streets
  • Future neighborhoods
  • Future transit routes

Current Pain Point

Walk
→ 1500 photos
→ Manually review 1500 photos
→ Select 207 archive photos
→ Generate project
→ Generate map
→ Generate stats
→ Generate downloads
→ Select 36 zine photos
→ Publish

The bottleneck:

1500 photos → 207 archive photos


Desired Workflow

Walk
→ Import photos to NAS
→ Mac Mini detects new project
→ Metadata extracted
→ Map generated
→ Stats generated
→ Download Project generated
→ AI proposes archive selection
→ Project created automatically
→ Dante reviews
→ Dante selects final 36
→ Publish

Goal:

When I arrive home, the project already exists.


Human Responsibilities

Focus on:

  • walking
  • photographing
  • exploring
  • observing
  • final zine editing

Avoid spending hours:

  • reviewing 1500 photos
  • building maps
  • calculating stats
  • creating project structures
  • uploading assets

AI Responsibilities

The AI should:

  • analyze photos
  • score photos
  • tag photos
  • identify duplicates
  • identify weak frames
  • identify strong frames
  • propose archive selections

The AI is learning:

“What does Dante include in the archive?”

Not:

“What is objectively good photography?”


Two Editing Layers

Layer 1 — Archive Edit

1500 photos
→ ~200 archive photos

Purpose: Preserve the walk.

Layer 2 — Zine Edit

~200 archive photos
→ 36 zine photos

For now, Dante performs this step manually.


Training Dataset

Every walk becomes a training example.

Example:

Ridge Ave

  • 1500 photos shot
  • 207 archive photos selected
  • 36 zine photos selected

Store:

  • all_photos/
  • archive_photos/
  • zine_photos/
  • metadata.json

Repeat for:

  • Market Street
  • Broad Street
  • Frankford Ave
  • Germantown Ave
  • Washington Ave
  • Ridge Ave
  • Future Philly walks

Sandbox Strategy

Do NOT start with 388,245 photos.

Start with:

  • Frankford
  • Market
  • Broad
  • Germantown
  • Washington
  • Ridge

Reason:

These projects already contain:

  • full walk
  • archive edit
  • zine edit
  • chronology
  • map data
  • metadata

Philly In Flux Hub

Create:

/philly-in-flux/

A dedicated command center for the project.


Master Statistics

Automatically update:

  • Total Photographs
  • Miles Walked
  • Steps Walked
  • Hours Walking
  • Hours Photographing
  • Projects Completed
  • Neighborhoods Reached
  • Days Documented
  • Archive Size

Philadelphia Progress System

Example:

PHILADELPHIA COVERAGE

██████░░░░░░░░░░░░░

18%

Track:

  • Market Street ✓
  • Broad Street ✓
  • Frankford Ave ✓
  • Germantown Ave ✓
  • Washington Ave ✓
  • Ridge Ave ✓
  • Walnut Street □
  • Chestnut Street □
  • Roosevelt Blvd □

Master Philadelphia Map

Features:

  • All completed walks
  • Route overlays
  • Street toggles
  • Project links
  • Coverage visualization

Users can:

  • toggle routes on/off
  • highlight specific streets
  • open projects directly

Project Archive

Every completed walk lives here.

Example:

Broad Street In Flux
161 photos
20.17 km

Market Street In Flux
115 photos
10.08 km


Timeline View

June 2026

  • Market Street
  • Germantown
  • Frankford
  • Washington
  • Ridge

Acts as a documentary logbook.


Automatic Project Generation

Given:

all_photos/

Generate:

  • metadata.json
  • route map
  • project statistics
  • project description
  • download project
  • contact sheet
  • archive selection proposal

Success Condition

Walk
→ Import
→ AI archive edit
→ Project auto-generated
→ Master Philadelphia statistics updated
→ Master Philadelphia map updated
→ Coverage percentage updated
→ Dante reviews
→ Dante selects final 36
→ Publish

Goal:

Remove repetitive labor.

Create more time for:

  • walking
  • photographing
  • exploring
  • documenting Philadelphia

while the archive grows continuously.

Scroll to Top