Talk:Using Imagery

From OpenStreetMap Wiki
Jump to navigation Jump to search

Opening

The opening three sentences of this page are a pretty poor way to welcome mappers that might only be doing remote mapping of places where there is no chance for local people to do on the ground surveys yet and suggests an ideology rather than an even handed, informative, helpful tone. I will be re-writing them when time permits. If someone would like to try and make them more even handed before then, it would be much appreciated. --Bgirardot (talk) 23:45, 8 January 2016 (UTC)

Agree, the wording was a bit too harsh. Jengelh made some good changes in the meantime. --Aseerel4c26 (talk) 13:23, 9 January 2016 (UTC)

Reorganisation

I'd propose reorganising this info into a different structure:

  1. Opener (which I've added) that quickly flags to newbies that aerial images are no panacea.
  2. Problems with aerial imagery
  3. Mitigation of those problems

This is in keeping with my feeling that the wiki too often tends towards overload on technical detail, rather than a progressive 'zooming in' so different levels of user can be familiarised slowly. Eteb3 (talk) 21:53, 15 September 2019 (UTC)

Terminology

Rather than 'Binding objects', would 'georeferencing' be a better term? That seems to be the term of art, but perhaps 'binding objects' is also? Eteb3 (talk) 21:55, 15 September 2019 (UTC)

Results of years of "Adjust imagery offset" use

Alas, years of 'Adjust imagery offset' use, by different, or even the same user, of different, or even the same imagery source, from different, or even the same date, often creates patches, most fitting in one screen, where the relative distances between items is correct, but there is discrepancy with the next group. I.e., many local datums!

Each user pulls the image to match the items already on the map, before editing. Alas each group did not start out the same, so the result is a patchwork of local datums.

Think about it, what else would you dare to do before adding a house? First pull the image so it lines up with the houses already on the map. Nobody can blame you. Sure, now it no longer lines up with the next village 500 meters away. But we all do our best... Jidanni (talk) 00:40, 8 October 2019 (UTC)

The only solution I can think of is to have some sort of artificial intelligence go in and from perfectly aligned imagery lay out a network of all the things it can.

People coming along later and editing would then naturally align the image they are using with these existing roads, rivers etc. before adding more features.

Sure one could say "we can't just put roads on the map via robot, not even knowing what Highway Number they are!"

Well, then have a whole new tag for these roads: "digital feature". Indeed, this "digital feature" tags could be used for everything automatically pulled from perfectly aligned images via robot. They would never be rendered, and only show up in the editor as a guideline to grow other features around. Jidanni (talk) 02:40, 8 October 2019 (UTC)

So the procedure would be, for some region:

  1. OSM acquires some one-time set of perfectly aligned imagery.
  2. Bot program is used to make a vector layer of all high contrast edges found in this imagery.
  3. When users enter the editor, they would notice this non-obtrusive grey etc. layer, which they naturally will drag the imagery they are using to align with. No, we don't want the grey to disappear, as they might edit somewhere else forgetting to see if the edges are still aligned.
  4. Voila, anything they add to the map in this area will now also be perfectly aligned, even with normal, non-perfectly aligned imagery!

OK I mentioned this topic on https://github.com/openstreetmap/iD/issues/1124#issuecomment-539291743 Jidanni (talk) 02:59, 8 October 2019 (UTC)