Hi all-

Praveen Ponnusamy, a savvy developer in our group, has cracked the Google Maps API so that single images can be used within Gmaps.  For places that store high-quality imagery in SDE or use Mr. Sid files, building 256 X 256 tiles on the fly is not very efficient.  It is much more efficient to just select one image over the entire vieawable window.   Praveen has worked through the Google javascript and has a working example of publishing Kansas orthophotography as a single image tile within the Google Maps framework.

http://geoportal.kgs.ku.edu/googlemaps/arcims_viewer.html

Praveen has also done a great write-up and supplied the code.  Read his write-up to find out more...

Single-tile version for publishing ArcIMS images using the Google Maps API.

http://geoportal.kgs.ku.edu/googlemaps/arcims_viewer.html

This is an example of using a "single tile" to publish ArcIMS images.  The viewer shows the default Google Map datasets together with city of Lawrence six-inch imagery and State of Kansas one-meter imagery.  The application uses and/or overrides undocumented methods of Google Maps code such as loadTileImagesLayer, configureImage, getBoundsBitmap, getLatLng and the much familiar getTileURL methods.  Also, due to namespace restrictions imposed by the Google javascript code, some of the utility methods referenced by the aforementioned methods had to be imported (copied) to the application code.

The application uses its own methods in place of the default loadTileImagesLayer and configureImage methods to publish the ArcIMS imagery and restores them back to the GMaps's implementation when the user switches to one of the default GMap types.  The overridden methods resemble their original implementation except for some minor changes to accommodate the single-tile architecture.  The trick to achieve this resides in the loadTileImagesLayer method where the for-loop to create the tiles is modified to create just one and to relocate it to the view window.  Finally, the moveend event is bound to query the image from server.  The server-side script called (to query the ArcIMS server) is the same as that used by Jeremy's tiled-version of ArcIMS viewer.

There are some gotchas to this approach though,

1. The lat-long values for the overlay and the tiles, for the view window, differ a little due to the difference in aspect ratios of the 256 x 256 images used by the GMaps and the variable image size (as dictated by the div element size, for example, 800 x 600) we use for the single-tile ArcIMS images.  The calculations for this adjustment are done in the getTileURL2 method of the new ArcIMS map specification class.  Essentially, the distance between the envelope's corners and the center of the view window, as given by the GMaps, is "scaled" and latitude-longitude values for the envelopes are recalculated.  This is important for the GMaps overlay (roads) and the ArcIMS image to be in sync.

2. Since undocumented methods are used here, the javascript code used by the application has to be updated whenever Google updates happen or else the application will break.  However, it should not take much time to update.

Future improvements
1. There may be a better/shorter way to do some of the stuff done here.
2. Cache images on-the-fly.
3. Increase the size of the images (possibly by 50%) queried from the server so as to have a buffer around the view window.
4. Add the ability to publish our own overlays too.

You can download the source code here: http://geoportal.kgs.ku.edu/googlemaps/GM_ArcIMS_1Tile.zip

ppraveen@kgs.ku.edu