When OpenStreetMap, Bing and other backgrounds (Portuguese photo) aren’t aligned, they are actually a little offset, which should I use as base reference?
Are you using iD or JOSM? In general, I would trust OSM to be relatively close to accurate for road centerlines, etc. Building footprints will almost always appear to be “off” from imagery due to errors in the footprint sources or oblique angles in the overhead imagery.
Others may have different advice, but I just try to make sure things are “correct” relative to a small area, for example, 4 street intersections defining a block. Even a very well-georeferenced map or overhead imagery can have inconsistent offsets in any sort of larger area. Does that make sense?
Also, in both iD and JOSM, you can fine tune the offsets as you move around a wider area using the imagery offset tools:
For iD:
For JOSM:
Does that help at all?
Aha! I think I’ve found at least one example that highlights your question:
Bing maps:
OSM:
So, it looks like my prior advice is relatively correct. I’d add just a reminder / request that we be sure not to trace anything from OSM. That may not be the case here, but we all need to be careful about that. The OSM way for the castle has been edited by 3 different OSM users, so we would need to get all of their permission to move this into OHM.
Looking at how they went about tagging it, on the other hand, can be quite helpful and informative. For example, the way that is used as the outer way of the castle relation in OSM, is also tagged with barrier=city_wall. Now, I’m not sure if that’s the correct or incorrect tag, but it is an interesting use of tagging relation members that adds meaningful descriptive information at the sub-relation level.
Here’s the Bing imagery aligned with the shape in OHM, using the image offset feature in iD. Once you’ve aligned the existing, mapped features, adding nearby features becomes much easier.
Hi @jeffmeyer
Thanks a lot for your time and tips!
1- When using iD, the two top backgrounds are marked with a *
This means something like preferred backgrounds?
2- the following screenshots show how (mis)aligned the feature is from the different photos. I know I can fix the alignment but let me explain my fears 
I used the most recent photos (ortophotos from 2021) and they were also probably used for OSM that’s the main (not the only but I’ll fix that reason why they are aligned: Ortophotos 2021, OSM and OHM.
The OHM feature is actually a little more detailed when compared to OSM and will be more as I’m planning to also use a plant drawing from portuguese heritage services.
Now, all the other photos (ortophotos 2018, bing and esri) show the castle a little bit to the north and if I go take a look at strees.gl it looks like they might be right because the 3D castle from OSM should be at the top, a little bit to the north.
I don’t know how accurate is the terrain information on streets.gl but all the other photo backgrounds (world imagery Waybacks) show a similar misalignment (photo show castle is a little bit to the north).
I know I can align the background but I’m not sure which photos should the base reference. In this case, I don’t have much references to decide if I should adjust the background or move the feature. It looks like I should move the feature…
Yes, the OSM community has determined that these layers are the “best”, better the default Bing aerial imagery in their coverage areas. “Best” can mean a variety of things – fresher, higher in resolution, less skewed or offset, taken during a time of year with less flooding, etc. Even if the layer is “best”, you may have good reasons for consulting other layers, because no layer is ever best in every respect.