[Imports] Thimphu City data import
Rafael Avila Coya
ravilacoya at gmail.com
Fri Jul 8 18:40:44 UTC 2016
Thank you for your comments on your experience.
Well, it saves not a few of nodes (even 17,000 in the case of 1.0)...
I see two options here:
1) No simplification beforehand. We leave that for the mappers when they
import those roads.
2) Checking carefully all roads. It would take me a lot, I guess. If I
feel I have enough energy, I will do it, but can't promise myself that
Is it ok?
On 08/07/16 14:59, Martin Koppenhoefer wrote:
> sent from a phone
>> Il giorno 08 lug 2016, alle ore 14:46, Rafael Avila Coya <ravilacoya at gmail.com> ha scritto:
>> Anyway: the simplification I did it with the simplify-way.max-error=3.0 (default). I did it now with that field at 1.0 (1 m máximum error), and it still reduces the number of nodes from 27,217 down to 10,579 , and the quality is (as expected) very equal to the original one. So I would change the streets file from the 3.0 simplified to the 1.0 simplified.
> I have also done some testing in the past on osm data (without uploading the results back) and have found that for most parts of the road network you won't notice worse data, but in some situations you will, especially in sharp curves and close to crossroads (if there is geometric detail in the shape one road enters the other this got lost or at least worse). Most benefit (and IMHO the only real application for OSM) is in overnoded, badly aligned data - obviously (because it doesn't matter in this case that you loose some detail, as this detail doesn't represent anything from reality). If the data is meticulously hand drawn (as opposed to automatic conversion from gpx data etc.) I would not do any automatic simplification as things will only get worse, less worse at 1m tolerance and more worse if you raise this level. Saving a few nodes doesn't justify the loss in accuracy (shape and position but mostly shape).
More information about the Imports