6 ms·

Rotate the World

76 points by jgrahamc 9y ago

pcunite 9y ago
When I was a kid, getting access to maps and globes was like a revelation. Now look at us, spinning something the size of the earth in web page. If you like looking at things from this orientation, give the celestiaproject.net a try.
nthitz 9y ago
Jason Davies' also did this with many more projections than mercator. Fun to browse https://www.jasondavies.com/maps/transition/ https://www.jasondavies.com/maps/transition/
- amelius 9y ago
  Fun indeed. It would be nice to know the advantages of each of these projections, and why they were developed.
  - jacobolus 9y ago
    Start with http://www.progonos.com/furuti/MapProj/Normal/TOC/cartTOC.html http://www.progonos.com/furuti/MapProj/Normal/TOC/cartTOC.ht... If you want formulas, https://pubs.er.usgs.gov/publication/pp1395 https://pubs.er.usgs.gov/publication/pp1395 If you want to know more, there are many books about cartography.
mikeash 9y ago
Are we ever going to be able to get rid of Mercator, or are we doomed to having people think that Greenland is as big as Africa until the end of time?
- njloof 9y ago
  Maybe on Big Block of Cheese Day? https://en.wikipedia.org/wiki/Gall–Peters_projection https://en.wikipedia.org/wiki/Gall–Peters_projection
  - clock_tower 9y ago
    Instead of swinging too far in the opposite direction, why not https://en.wikipedia.org/wiki/Robinson_projection https://en.wikipedia.org/wiki/Robinson_projection ? The northern latitudes are bigger in Robinson's projection than they are in real life, but you do need to refer to them pretty often, and they're not ridiculous like they are in Mercator. (On further reading: the https://en.wikipedia.org/wiki/Winkel_tripel_projection https://en.wikipedia.org/wiki/Winkel_tripel_projection looks a bit more 'stretched', but keeps the northern latitudes readable while not exaggerating their size.)
- jacobolus 9y ago
  Here’s my suggested alternative, also conformal, http://i.imgur.com/Y6ki0l9.jpg http://i.imgur.com/Y6ki0l9.jpg Inspired by http://www.genekeyes.com/Cahill-desk-maps/Cahill-1918-rotated-25pc.jpg http://www.genekeyes.com/Cahill-desk-maps/Cahill-1918-rotate...
  - mikeash 9y ago
    That seems nice for paper. I'd like to see actual spheres used for computers. We don't have to stick with fixed 2D projections for electronics. Render the Earth as a sphere centered on the center of the viewport. When zoomed in, it'll look flat (just as the Earth itself looks flat up close) but without the size distortions you get when you scroll e.g. Google Maps far north and south. Zoomed out, you'll see how everything actually relates in 3D space.
    - jacobolus 9y ago
      By “actual spheres” what you presumably mean is an orthographic projection (or maybe some kind of perspective projection). But such projections are almost never the best choice for any purpose, unless you’re trying to make a game about flying a little rocket around the solar system or something. As maps, they both hide significant parts of the globe and dramatically distort the rest. The point of a map is usually not to just be a stand-in for a hand-held globe, but to help a viewer understand and answer questions about geography. The zoomed out views of Google Earth etc. are mostly gimmicks.
      - mikeash 9y ago
        I know these projections would have big problems if they were static, but it seems to me that the whole game changes when it's dynamic. The fact that half the globe is hidden from view is much less important when a flick of a finger can bring it into view. I admit I don't quite get that bit about not just being a stand-in for a globe. I thought the main reasons maps aren't globes is just because maps are much easier to store, and because maps can be zoomed in to an arbitrary degree. Computer displays eliminate both of those reasons.
- rspeer 9y ago
  You shouldn't want to actually "get rid" of Mercator entirely. Mercator is pretty great for interactive maps on the Web. Conformal angles, north is always up, splits nicely into tiles. As long as Africa and Greenland aren't on the same map, it's got almost everything that you want from a map. Of course, what's best for an interactive map is if it turns into a globe when you zoom out, like Google Earth. instead of being a shitty global Mercator. That's a lot more graphically complex, though.
  - mikeash 9y ago
    Rendering a 3D globe onto a 2D screen, centered on the current viewing area and with a fixed orientation, essentially gets what you describe. When you're zoomed in you see something that very closely approximates a Mercator projection cropped to that area. You'll get conformal angles and North is always up. Tiling probably isn't as easy, but it's a Simple Matter of Programming. As you zoom out, it gradually and seamlessly transitions from Mercator-like to globe-like.
  - robbrown451 9y ago
    Why not have it stay as a globe when you are zoomed in? It will be similar to Mercator, but the aspect ratio would be more accurate especially as you get further from the equator. Admittedly, it is easier for developers who are adding features if you do it as Mercator, but not that much easier (if it has a good API).
    - rspeer 9y ago
      I don't know what you mean about the aspect ratio further from the equator. Mercator is conformal, it preserves angles, so the aspect ratio is 1:1 everywhere that it makes sense to say "aspect ratio". Are you talking about the problem with the shape of a Mercator tile becoming absurd very close to the poles, which I wouldn't call "aspect"? Or are you thinking of the equirectangular projection, where the aspect ratio changes to preserve the distance between latitude lines? The reason not to use a globe is efficiency. Not only is 2D easier than 3D, but you can make Mercator out of tesselated tiles. You can't do that on a globe without having way too many tiles or obvious seams.
mrgriscom 9y ago
Also this: http://mrgris.com/projects/merc-extreme/ http://mrgris.com/projects/merc-extreme/
- jacobolus 9y ago
  For more along these lines, http://archive.bridgesmathart.org/2013/bridges2013-217.pdf http://archive.bridgesmathart.org/2013/bridges2013-217.pdf
gedy 9y ago
This is wonderful, some of the first programming I did was to try and do this in GWBASIC on a 286 many years ago. It's magical to see this in realtime in your browser.

76 points by jgrahamc 9y ago

When I was a kid, getting access to maps and globes was like a revelation. Now look at us, spinning something the size of the earth in web page. If you like looking at things from this orientation, give the celestiaproject.net a try.

nthitz 9y ago

Jason Davies' also did this with many more projections than mercator. Fun to browse https://www.jasondavies.com/maps/transition/ https://www.jasondavies.com/maps/transition/

amelius 9y ago
Fun indeed. It would be nice to know the advantages of each of these projections, and why they were developed.
- jacobolus 9y ago
  Start with http://www.progonos.com/furuti/MapProj/Normal/TOC/cartTOC.html http://www.progonos.com/furuti/MapProj/Normal/TOC/cartTOC.ht... If you want formulas, https://pubs.er.usgs.gov/publication/pp1395 https://pubs.er.usgs.gov/publication/pp1395 If you want to know more, there are many books about cartography.

mikeash 9y ago

Are we ever going to be able to get rid of Mercator, or are we doomed to having people think that Greenland is as big as Africa until the end of time?

njloof 9y ago
Maybe on Big Block of Cheese Day? https://en.wikipedia.org/wiki/Gall–Peters_projection https://en.wikipedia.org/wiki/Gall–Peters_projection
- clock_tower 9y ago
  Instead of swinging too far in the opposite direction, why not https://en.wikipedia.org/wiki/Robinson_projection https://en.wikipedia.org/wiki/Robinson_projection ? The northern latitudes are bigger in Robinson's projection than they are in real life, but you do need to refer to them pretty often, and they're not ridiculous like they are in Mercator. (On further reading: the https://en.wikipedia.org/wiki/Winkel_tripel_projection https://en.wikipedia.org/wiki/Winkel_tripel_projection looks a bit more 'stretched', but keeps the northern latitudes readable while not exaggerating their size.)
jacobolus 9y ago
Here’s my suggested alternative, also conformal, http://i.imgur.com/Y6ki0l9.jpg http://i.imgur.com/Y6ki0l9.jpg Inspired by http://www.genekeyes.com/Cahill-desk-maps/Cahill-1918-rotated-25pc.jpg http://www.genekeyes.com/Cahill-desk-maps/Cahill-1918-rotate...
- mikeash 9y ago
  That seems nice for paper. I'd like to see actual spheres used for computers. We don't have to stick with fixed 2D projections for electronics. Render the Earth as a sphere centered on the center of the viewport. When zoomed in, it'll look flat (just as the Earth itself looks flat up close) but without the size distortions you get when you scroll e.g. Google Maps far north and south. Zoomed out, you'll see how everything actually relates in 3D space.
  - jacobolus 9y ago
    By “actual spheres” what you presumably mean is an orthographic projection (or maybe some kind of perspective projection). But such projections are almost never the best choice for any purpose, unless you’re trying to make a game about flying a little rocket around the solar system or something. As maps, they both hide significant parts of the globe and dramatically distort the rest. The point of a map is usually not to just be a stand-in for a hand-held globe, but to help a viewer understand and answer questions about geography. The zoomed out views of Google Earth etc. are mostly gimmicks.
    - mikeash 9y ago
      I know these projections would have big problems if they were static, but it seems to me that the whole game changes when it's dynamic. The fact that half the globe is hidden from view is much less important when a flick of a finger can bring it into view. I admit I don't quite get that bit about not just being a stand-in for a globe. I thought the main reasons maps aren't globes is just because maps are much easier to store, and because maps can be zoomed in to an arbitrary degree. Computer displays eliminate both of those reasons.
rspeer 9y ago
You shouldn't want to actually "get rid" of Mercator entirely. Mercator is pretty great for interactive maps on the Web. Conformal angles, north is always up, splits nicely into tiles. As long as Africa and Greenland aren't on the same map, it's got almost everything that you want from a map. Of course, what's best for an interactive map is if it turns into a globe when you zoom out, like Google Earth. instead of being a shitty global Mercator. That's a lot more graphically complex, though.
- mikeash 9y ago
  Rendering a 3D globe onto a 2D screen, centered on the current viewing area and with a fixed orientation, essentially gets what you describe. When you're zoomed in you see something that very closely approximates a Mercator projection cropped to that area. You'll get conformal angles and North is always up. Tiling probably isn't as easy, but it's a Simple Matter of Programming. As you zoom out, it gradually and seamlessly transitions from Mercator-like to globe-like.
- robbrown451 9y ago
  Why not have it stay as a globe when you are zoomed in? It will be similar to Mercator, but the aspect ratio would be more accurate especially as you get further from the equator. Admittedly, it is easier for developers who are adding features if you do it as Mercator, but not that much easier (if it has a good API).
  - rspeer 9y ago
    I don't know what you mean about the aspect ratio further from the equator. Mercator is conformal, it preserves angles, so the aspect ratio is 1:1 everywhere that it makes sense to say "aspect ratio". Are you talking about the problem with the shape of a Mercator tile becoming absurd very close to the poles, which I wouldn't call "aspect"? Or are you thinking of the equirectangular projection, where the aspect ratio changes to preserve the distance between latitude lines? The reason not to use a globe is efficiency. Not only is 2D easier than 3D, but you can make Mercator out of tesselated tiles. You can't do that on a globe without having way too many tiles or obvious seams.

mrgriscom 9y ago

Also this: http://mrgris.com/projects/merc-extreme/ http://mrgris.com/projects/merc-extreme/

jacobolus 9y ago
For more along these lines, http://archive.bridgesmathart.org/2013/bridges2013-217.pdf http://archive.bridgesmathart.org/2013/bridges2013-217.pdf

gedy 9y ago

This is wonderful, some of the first programming I did was to try and do this in GWBASIC on a 286 many years ago. It's magical to see this in realtime in your browser.