Best Imagery Lookup

[bhousel]

I am proposing...

A new database and service to track which imagery is best and what adjustments to apply to it to make it look amazing.

What will this enable that we can't do today?

Some of the biggest things that I see new mappers struggle with are:

• knowing which imagery to choose
• knowing to how/whether to apply an imagery offset

This problem of choosing the best imagery actually overwhelms people. I've seen them totally panic at mapathons! We don't want this.

What exists today:

editor-imagery-index is our list of imagery that we are allowed to trace from in the OSM editors. Each source can optionally contain a best: true property to say "it's the best".

This solution was ok back when the only sources to choose from were "Bing" and "Mapbox" and whatever local imagery as available. However we now have 2 more DigitalGlobe sources, 2 more ESRI sources, and many more local sources. Each of these will be great in some places and bad in others.
This is a great problem to have!

We've written a bit about this on osmlab/editor-layer-index#130

We also have Imagery Offset Database, built by @Zverik. It's pretty good and has a lot of usage and a JOSM plugin which can auto-adjust the imagery. However I'd like something that does a bit more than just offsets, and understands actual tiles, not lat/lngs. This database can not tell you if Bing at z19 is good and Bing at z17 is bad - and the answer to that question changes all the time.

How

I feel like the ideal solution is going to be tile-based, not polygon-based:

For tile z/x/y (or quadkey nnnnnn) , give me the id of the best imagery, the offset to apply to it, and any enhancements that might help it look better

("enhancements" means, brightness, contract, saturation, sharpness from openstreetmap/iD#4575)

Whom does this benefit?

Mostly beginner users, but also advanced users who want to save time, and mappers who don't want to see their edits broken by other users looking at bad imagery.

Any drawbacks?

It would be another thing to maintain.

Resources needed to build and maintain?

I'd like to see this run as an independent project (in other words, not something to add to the OSM database itself).

See also:

• editor-imagery-index
• Imagery Offset Database is good

Any ideas for a project name?

🤷‍♂️

Describe this project in a single emoji:

🖼

[mcawley]

Ideal: Here's the best imagery layer, here is the offset, and here are the display options applied.

[bhousel]

Good usecase for this: Mapathon organizer wants to set up the imagery preferences in an area so that everyone attending does not need to do it.

As a good first step, implement saving this info in iD to user's local storage in JSON blob, and figure out later how to persist these to a server (stored on S3 named quadkey files?)

Unsolved:

• what happens when multiple people apply preferences (conflicts?)

[Marc-marc-marc]

any comment from @Zverik to known if an extend for Imagery Offset Database can fill the need ?

[1ec5]

Often, there is no single “best” layer: one layer is more up-to-date, another higher-resolution, yet another taken in the fall so you can see through trees. I’m unsure which of these attributes is important enough that it should determine the default, but an armchair mapper needs to be able to cycle effortlessly among them. I’d love to be able to encode this table in such a way that an editor like iD can offer a choice of:

• ( ) Current
• ( ) High-Resolution
• ( ) Well-Aligned
• ( ) Leaf-Off

The choices would need to vary by region and zoom level, to address the mosaic issue mentioned in the original post.