Wiki page: Roof modelling

Being quite inexperienced in OSM, I am trying to learn more about mapping techniques. While doing so, I noticed an interpretation on the following Wiki page that I think to be incorrect. So I just wanted to draw somebody’s attention to it:
states that:
Due to perspective, the roof is not necessarily the same size as the base, and the polygon may need to be enlarged/shrunk a little.

Given the far distance from the satellite, I would not think it possible that the perspective could change as much as that in relation to the small area on the image. It appears to me that the reason for the size difference is not the perspective, but the height of the second building. I conclude this from the visible wall area as well as the length of the shadows of the two buildings. I would also suspect that the two buildings are not joined together, but separated by a walk-through.

My conclusion would therefore be that the two roofs should be separately outlined and separately positioned.

Having said all that, I am not even thinking about getting involved in the Wiki pages. I’m already having a hard time getting my head around the OSM. But maybe somebody that is already involved may want to take a look at this. If I am right, that means the information on that page is wrong. And that could cause even more confusion for people like myself trying to get the hang of editing.


We mostly map with aerial imagery rather than satellite data, and with aerial images the perspective does matter.

Yes, we do, but

is true nonetheless for the example given in the wiki. The buildings are of different height.

I am unsure about your answer. Isn’t the aerial imagery (i.e. Bing) taken by a satellite?

no, mostly not, the higher resolution imagery we are typically using is recorded from airplanes

1 Like

@ dieterdreist
Thanks, one learns something new every day.

1 Like

Just to confirm Nadjita - do you agree with my observation that the marking of the second building in the wiki would be incorrect?
Maybe I should clarify that I am referring to the length of the building being extended and there not being a space between the two buildings.

Yes. It’s also clearly visible that the length of the shadows of both buildings is different. I’m not 100% sure, but it looks liek they don’t touch each other.

Update: if you use bird’s eye view, or updated imagery from bing, it confirms this:


The example shows that the perspective distortion between the (flat) roof and the base area (here for simple cuboid buildings) plays a smaller role than assumed. The author is subject to a different problem/assumption of perspective and has made a mistake. If so, then the higher roof should be reduced in size, not enlarged.

@Lithbeth well recognised that something cannot be right here. And @Nadjita did excellent research and found the actual error.

:+1: :+1: :+1: