osmlab/labuildings

Data offset

Closed this issue · 2 comments

From the latest sample output by @maning #14 (comment)

I'm noticing a slight but consistent offset of the data from the background of the order of 2m with both Bing and Mapbox imagery.

untitled2

GPS information is not accurate enough to verify the offset to this level. Should we ignore this and consider the data to be correct?

@almccon @maning

Using a custom -s_srs parameter in ogr2ogr, I'm getting better results with the LA City 2014 building data #32. See notes here: https://gist.github.com/maning/093a866fd465ea880032#better-conversion

Next steps
Run a test convert to all data layers (blocks, addresses, 2008 buildings) then, change reprojection script in the Makefile.

@maning the sample data matches Bing almost perfectly, and the offset with Mapbox imagery has reduced.

untitled2

The projection looks good 👍