Workflow for the National Geographic World Atlas Mobile App

Rosemary Wardley, National Geographic Maps
Steve Gifford, Mousebird Mobile
#nacis2015

Transcript

1. Data Processing Workflow
   for the
   National Geographic
   World Atlas Mobile App
   Rosemary Wardley
   Steve Gifford
   

   View Slide

2. Introduction
   World Atlas App
   Previous versions & Problems
   Goals for new app
   Data production
   Data processing
   Conclusions
   

   View Slide

3. World Atlas App for iPhone/iPad
   Digital & Interactive Atlas
   Political basemaps
   Flags & Facts
   Search Index
   Statistical Overlays
   Distance Calculator
   

   View Slide

4. How  it  was  done  before
   

   View Slide

5. How  it  was  done  before
   

   View Slide

6. How  it  was  done  before
   

   View Slide

7. How  it  was  done  before
   

   View Slide

8. How  it  was  done  before
   

   View Slide

9. App  Design/UX  Flaws
   

   View Slide

10. App  Design/UX  Flaws
    

    View Slide

11. App  Design/UX  Flaws
    

    View Slide

12. Goals  for  World  Atlas  v.4
    Tighter  control  over  the  data  for  editorial  purposes  
    Leverage  more  of  our  databases  
    Reduce  the  amount  of  custom  processing  
    Fit  into  evolving  Editorial/Production  workflow  
    App  data  that  is  easier  to  update  &  replicate  
    Maintain  NG  brand  identity    
    Add  flexibility  &  reduce  size  of  the  App  
    Hybrid  Raster  &  Vector  process
    

    View Slide

13. Full  Workflow
    

    View Slide

14. Data  Production
    

    View Slide

15. Inputs:  Standard  Carto  Datasets
    75  Million  
    Zooms  0-­‐2
    40  Million  
    Zooms  2-­‐4
    20  Million  
    Zooms  5-­‐6
    10  Million  
    Zooms  7-­‐8
    

    View Slide

16. Editorial  &  Production  Steps
    

    View Slide

17. GFID  match
    

    View Slide

18. Flags  &  Facts
    

    View Slide

19. Search  Index  Creation
    

    View Slide

20. GIS  Database  Steps
    

    View Slide

21. Final  Product
    

    View Slide

22. Raster  &  Vector  Outputs
    Background  
    Raster
    Full  PDF
    Symbol  
    Shapefiles
    Boundary  
    Shapefiles
    

    View Slide

23. Data  Processing
    

    View Slide

24. Why Vectors?
    Image Artifacts
    Blurry text & symbols
    Tile edge discontinuities
    Size
    Images are big
    Vectors are small
    Selection Data
    Tapping logic simpler
    

    View Slide

25. Vector Processing
    Symbols
    Points from GIS database
    Both visual and selection
    Text
    Extracted from PDF
    Purely visual
    Boundaries
    Areals from GIS database
    Invisible, just for selection
    

    View Slide

26. Background Raster
    Including
    Roads, Rivers
    Fills, stipples
    Etc (lots of etc)
    Excluding
    Symbols
    Text
    

    View Slide

27. Text Extraction
    Parse the PDF
    Need for each character
    Location
    Rotation
    Size
    Font
    Output to Shapefile
    

    View Slide

28. Uploading to Mapbox
    Images basemaps
    Low res everything
    High res blanks
    Vectors
    Symbols
    Text
    Boundaries
    

    View Slide

29. Anatomy of a Tile
    

    View Slide

30. How did it go?
    App is released and working
    Text is Sharp
    Symbols are Sharp
    Data size is way down
    Old app: 500MB per layer
    New app:75MB per layer
    

    View Slide

31. Conclusions:  Data  Update  Process
    

    View Slide

32. Conclusions:  Software  Used
    Mapbox Studio
    

    View Slide

33. View Slide

34. Wrapup & Contact
    Rosemary Wardley
    rwardley@ngs.org
    nationalgeographic.com
    Steve Gifford
    sjg@mousebirdconsulting.com
    mousebird.github.io/WhirlyGlobe/
    

    View Slide