cfis : Google Maps Deconstructed

Comment on Google Maps Deconstructed by Brian Flood

Wed, 03 May 2006 07:33:37 -0600

Great post!

Comment on Google Maps Deconstructed by Charlie

Wed, 03 May 2006 11:37:58 -0600

Thanks Brian. Think your blog is great by the way - I enjoy reading it.

Comment on Google Maps Deconstructed by p3n

Fri, 05 May 2006 09:53:58 -0600

thx

Comment on Google Maps Deconstructed by gisdudester

Mon, 08 May 2006 08:11:22 -0600

this was very informative. considering I have a certificate in GIS, i still get Geographic Coordinates Systems and Projected Coordinate Systems mixed up. but these are very logical and progress in an intuitive way.

Comment on Google Maps Deconstructed by Paul

Fri, 30 Jun 2006 12:10:57 -0600

Hi Charlie,

I think your wgs84 assumption is accurate. I have reproduced Google’s top level tile (world in 256x256) using mapserver and this proj.4 definition:

World Mercator +proj=merc +lat_ts=0 +lon_0=0 +k=1.000000 +x_0=0 +y_0=0 +ellps=WGS84 +datum=WGS84 +units=m no_defs <>

and with the following extents:

-20037508.342789, -19994875.249796, 20037508.342789, 19994875.249796

which is reprojected from the lonlat extents -180,-85.05112877980660,180,85.05112877980660

The image produced is essentially identical (less antialiasing and taking some generalization into account).

Cheers

Paul

Comment on Google Maps Deconstructed by Paul

Fri, 30 Jun 2006 12:15:18 -0600

Hi Paul,

That’s great, thanks for the verification. I suspected it was true for the US since I can take GPS tracks and overlay them perfectly ontop of Google Maps (its pretty cool).

What data are you generating the tiles from - Tiger?

Trackback from cfis: Mouse Coordinates To Lat/Long on Google Maps Deconstructed

Fri, 30 Jun 2006 16:45:11 -0600

This post is long overdue - a number of people have asked me how to convert between lat/long and mouse coordinates on an HTML page. First, let's define some terms since otherwise is easy to get confused: Geodetic coordinate system- A geodetic coor...

Comment on Google Maps Deconstructed by Nikolas Coukouma

Fri, 07 Jul 2006 17:56:26 -0600

The reason Google Maps uses the Mercator projection isn’t because it’s easy to implement. One of the major features, arguably the primary one, of GMaps is providing directions from point A to point B. Usability studies showed that the angle of turns is very important, much more than displayed distance. The Mercator projection preserves angles.

I picked this information up from Bret Taylor, who gave a tech talk on GMaps during my visit to Google two weeks ago.

Comment on Google Maps Deconstructed by Charlie

Sun, 09 Jul 2006 02:16:41 -0600

Hi Nikolas,

Thanks for the comment - that’s really interesting. Makes me curious how far you’d have to be zoomed out (i.e, how far are you travelling) for the angles to be distorted - have to check that out.

Google could have used other conformal projections (angle preserving) - I seems to me that the Mercator projection does make implementation easier. Was there any mention of that in the talk?

Comment on Google Maps Deconstructed by James

Wed, 20 Feb 2008 13:26:09 -0700

Hi There,

Does anyone know how to calculate the bitmap x,y cordinates in a tile given latitude and longitude?

Cheers!

James

Comment on Google Maps Deconstructed by naimulah

Thu, 20 Mar 2008 11:13:22 -0600

very informative and interesting article you have written. Thanks Charlie.

Keep up the good work.

Comment on Google Maps Deconstructed by Smurf

Sat, 21 Jun 2008 01:09:12 -0600

Mercator doresn’t distort angles as you zoom out; it bends straight lines (i.e. great circles) instead.

They could have used a Mercator projection based on the center of your screen, but that’d require a new set of tiles for every PoV, or (assuming a current large screen with room for 58 tiles), 40 times the computing power required to render these things, and 40 times the storage requirement. Which is already huge*.

They get around the problem in Google Earth by distorting the tiles on the client machine, but you can’t do that in a web server.

Trackback from SharpGIS: The Microsoft Live Maps and Google Maps projection on Google Maps Deconstructed

Sun, 10 Aug 2008 04:14:46 -0600

The Microsoft Live Maps and Google Maps projection

Comment on Google Maps Deconstructed by Klokan Petr Přidal

Mon, 18 Aug 2008 10:11:30 -0600

The coordinate system and projection is described at MapTiler’s Google Maps Projection and Coordinate system site.

There is also source code for script implementing all the math for coordinate conversion. I think it is worth to look at for readers of this blog post.

Comment on Google Maps Deconstructed by tz

Wed, 22 Oct 2008 07:24:14 -0600

One problem. The maps at the higher zoom levels are not horribly distored.

Case in point, Barrow Alaska, 71.3 lat, -156.8 lon.

If you look at the images and other things which are supposed to be square, they are square. cos(70) is under .5 so they should be badly stretched sideways if it was true Mercator.

This is also true of cities and counties laid out on a grid (a street every mile) - they look square and not like rectangles or rhombuses. Mercator projections of the states near the Canadian border should have this very visible but are not.

I assume a 1-mile north = 1 mile east map of barrow (or MI or MN) looks like the image on google maps and google earth (whose ruler says 1 mile for the same distance north or east).

Something is wrong, or at the highest zooms something more is being done. It all works well, but is not a mercator projection at some point.

Comment on Google Maps Deconstructed by cc

Sat, 15 Nov 2008 16:13:57 -0700

great post, thanks!