On the one hand, there’s a pretty widespread principle that an individual raw fact is ineligible for copyright protection on its own. On the other hand, I wouldn’t go so far as to say that “extracting” details from a photo avoids copyright considerations.
Tracing roads and buildings from another map is definitely copying. Tracing the same roads and buildings from an aerial imagery layer is also copying. If you instead trace them from a photo taken from a hot air balloon, it would still be copying. If the photo shows the building’s wall because the hot air balloon hasn’t taken off yet, that’s still copying. If you have the photo side by side instead of overlaid, that’s still copying. If you’ve printed out the photo and refer to it on your desk while mapping… you get the idea.
In some jurisdictions, all these conceptual hops might add up to a fair use defense against a claim of copyright infringement, or even a determination that there wasn’t any copyright infringement to begin with. However, OSM is based in the UK, where there historically haven’t been similarly strong protections for reusers. And as a matter of project policy, we don’t want to rely on legal gray areas anyways.
This can be frustrating sometimes, since most of the authors you’d need to reach out to don’t have such a rigorous, pedantic stance on intellectual property rights. I’ve all but angered some local government agencies by pestering them about the copyright status of works they don’t really want to protect in the first place (but can’t say so explicitly). How ironic that it falls on an open data project to educate others about the importance of respecting legal barriers around data!
If you can, maybe ask the Swetrails user if they’d be willing to let you put this information in OSM without any restrictions, perhaps crediting them less formally in source=* tags or somewhere on the wiki. It’s easier for everyone than getting into the weeds of formal licenses and waivers. Like OSM, most crowdsourced sites have terms that allow contributors to retain ownership of their contributions, so they’d have the right to give you that permission.
Wikimedia Commons has different considerations due to a difference in focus, use case, and jurisdiction. Most of the coordinates on image description pages come from EXIF data encoded in the image, but some are manually entered more manually by Commons contributors based on their own geolocation efforts. More recently, Commons has started putting this information in structured form that’s explicitly dedicated to the public domain under CC0. Sometimes I use these coordinates to jump to the general location in OSM, knowing that EXIF data is only as reliable as a single stray GPS reading, and then I use my normal mapping skills to determine the actual location.
Correct, it doesn’t matter if you’re copying from Google’s basemap or their aerial imagery. It’s all copyrighted. Neither their terms of service nor OSM policy allows you to copy from either layer of theirs.