On Wednesday 04 July 2018, Frederik Ramm wrote: > > What you are suspecting here seems to have manifested itself in > Alaska, where one reviewer claims that most (!) buildings in the > dataset do not correspond to anything on the imagery available to us, > and the few (!) objects that do differ wildly in shape:
I also thought about Alaska but i was trying to give Microsoft the benefit of the doubt here and not make the assumption that any quality issues in Alaska would allow conclusions for the rest of the US. Alaska is not only very different in geography, it is also much more difficult with images (no NAIP images, no full HR coverage in any of the common image layers). For Alaska is is also highly significant what coordinate system Microsoft did the processing in. If they used Mercator and did not take into account the scale variation this is very likely to be falling onto their feet obviously. Even for just CONUS you by the way have a scale factor variation of about 1.35 which could already cause problems when not compensated for. But now i am drifting off-topic here myself... ;-) -- Christoph Hormann http://www.imagico.de/ _______________________________________________ Imports mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports
