[OpenLayers-Users] ArcGIS Online Cached tiles datum shift
Vadim Bokin
vbokin at gmail.com
Tue Jul 20 02:20:58 EDT 2010
Ok, now I'm confused. How do I initialize the ArcGISCache layer with
this service
http://server.arcgisonline.com/ArcGIS/rest/services/World_Street_Map/MapServer
??
Previously I was going after
http://server.arcgisonline.com/ArcGIS/rest/services/World_Street_Map/MapServer
+ "/tile/" + "${z}/${y}/${x}" tiles.
This goes after completely different scheme of tiles.
On Mon, Jul 19, 2010 at 9:08 PM, Mike Swope <mswope at gmail.com> wrote:
> You might also keep in mind, that they are using arcgis.com now. Perhaps
> some more info is there?
>
> On Mon, Jul 19, 2010 at 9:17 PM, Vadim Bokin <vbokin at gmail.com> wrote:
>>
>> Robert,
>>
>> I've made good progress with your suggestion. I created a new class
>> and just extended XYZ.js with the 2 methods you provided. But I still
>> have a problem.
>>
>> Old Code: locations where close, but kept jumping around north or
>> south of the right location
>>
>> New code: locations are almost exactly 41.0 miles North of where they
>> should be. But they are off by a fixed amount. So when I zoom in and
>> out they are staying put where they are, but they are still off. So
>> if I try to put a point on the beach in Santa Monica, on the map it
>> looks to be between Santa Clarita and Palmdale. As soon as I change
>> base maps to Google or a epsg:4326 map, all is well and points are in
>> perfect location.
>>
>> So it's 1 step forward (locations don't jump) and 1 step back
>> (locations are pretty far north).
>>
>> I read tileOrigin from the AGS layer spec, e.g.
>>
>> http://server.arcgisonline.com/ArcGIS/rest/services/World_Street_Map/MapServer.
>> And I've made sure it is correct by testing in firebug and just
>> dumping out map.baseLayer.tileOrigin.
>>
>> Did you do anything else to modify XYZ to make this work? Does 41.0
>> miles mean anything? to me it doesn't look like 1 degree, or anything
>> else. But it's a fixed distance difference between plotted and viewed
>> points. Some kind of error in calculation or my setup of the layer?
>>
>> Thank you!
>>
>> -Vadim
>>
>>
>> On Mon, Jul 19, 2010 at 6:24 PM, Robert Sanson
>> <Robert.Sanson at asurequality.com> wrote:
>> > Hi Vadim
>> >
>> > What layer type are you using? ArcGIS tilecaches have a tile origin in
>> > the
>> > top left. The standard layer.XYZ in OpenLayers is built to support a
>> > lower
>> > left tile origin. You need a patched XYZ.js with setMap and
>> > calculateGridLayout copied from Zoomify.js:
>> >
>> > * APIMethod: setMap
>> > * When the layer is added to a map, then we can fetch our origin
>> > * (if we don't have one.)
>> > *
>> > * Parameters:
>> > * map - {<OpenLayers.Map>}
>> > */
>> > setMap: function(map) {
>> > OpenLayers.Layer.Grid.prototype.setMap.apply(this, arguments);
>> > if (!this.tileOrigin) {
>> > this.tileOrigin = new OpenLayers.LonLat(this.maxExtent.left,
>> >
>> > //this.maxExtent.bottom);
>> > this.map.maxExtent.top);
>> > //patched to be similar to Zoomify
>> > }
>> > },
>> >
>> > /**
>> > * Method: calculateGridLayout //patched from Zoomify
>> > * Generate parameters for the grid layout. This
>> > *
>> > * Parameters:
>> > * bounds - {<OpenLayers.Bound>}
>> > * extent - {<OpenLayers.Bounds>}
>> > * resolution - {Number}
>> > *
>> > * Returns:
>> > * Object containing properties tilelon, tilelat, tileoffsetlat,
>> > * tileoffsetlat, tileoffsetx, tileoffsety
>> > */
>> > calculateGridLayout: function(bounds, extent, resolution) {
>> > var tilelon = resolution * this.tileSize.w;
>> > var tilelat = resolution * this.tileSize.h;
>> >
>> > var offsetlon = bounds.left - extent.left;
>> > var tilecol = Math.floor(offsetlon/tilelon) - this.buffer;
>> > var tilecolremain = offsetlon/tilelon - tilecol;
>> > var tileoffsetx = -tilecolremain * this.tileSize.w;
>> > var tileoffsetlon = extent.left + tilecol * tilelon;
>> >
>> > var offsetlat = extent.top - bounds.top + tilelat;
>> > var tilerow = Math.floor(offsetlat/tilelat) - this.buffer;
>> > var tilerowremain = tilerow - offsetlat/tilelat;
>> > var tileoffsety = tilerowremain * this.tileSize.h;
>> > var tileoffsetlat = extent.top - tilelat*tilerow;
>> >
>> > return {
>> > tilelon: tilelon, tilelat: tilelat,
>> > tileoffsetlon: tileoffsetlon, tileoffsetlat: tileoffsetlat,
>> > tileoffsetx: tileoffsetx, tileoffsety: tileoffsety
>> > };
>> > },
>> >
>> > Using the above works perfectly.
>> >
>> > regards,
>> >
>> > Robert Sanson
>> >
>> >>>> Vadim Bokin <vbokin at gmail.com> 20/07/2010 11:53 a.m. >>>
>> >
>> > Hi,
>> >
>> > Does anyone have the _exact_ proj4js definition for ESRI projections
>> > EPSG:102100 and the older EPSG:102113?
>> >
>> > My maps have a y-coordinate shift (both north and south at different
>> > zooms)
>> > off by many map pixels. Always close, but no cigar. X-coordinate is
>> > always
>> > dead on.
>> >
>> > I can add in ArcGIS online content as dynamic layer without any
>> > problems.
>> > But when adding cached tiles, there's always this shift. Has anyone
>> > seen
>> > this before and/or been able to resolve it? I need to be able to use
>> > the
>> > cached tiles.
>> >
>> > I've used the one from spatialreference.org, as well as set it directly
>> > to
>> > be equal to Google's EPSG:900913 but there's always a shift. I've tried
>> > many different variations with +a and +rf and others but I feel like
>> > the
>> > definition is missing a +towgs84=... adjustment to properly correct for
>> > the
>> > shift on the map but I have no way of knowing what those parameters
>> > should
>> > be and whether it should be the 3-parameter adjustment or the
>> > 7-parameter
>> > one.
>> >
>> > Here's my current definition:
>> >
>> > Proj4js.defs["EPSG:102100"] = "+proj=merc +lon_0=0 +x_0=0 +y_0=0
>> > +a=6378137
>> > +b=6378137 +units=m +nadgrids=@null";
>> >
>> > +towgs84=0,0,0,0,0,0,0 doesn't help, but I feel like the right values
>> > in
>> > there will fix the problem.
>> >
>> > Thank you,
>> >
>> > -Vadim
>> >
>> >
>> > --
>> > View this message in context:
>> >
>> > http://osgeo-org.1803224.n2.nabble.com/ArcGIS-Online-Cached-tiles-datum-shift-tp5314596p5314596.html
>> > Sent from the OpenLayers Users mailing list archive at Nabble.com.
>> > _______________________________________________
>> > Users mailing list
>> > Users at openlayers.org
>> > http://openlayers.org/mailman/listinfo/users
>> >
>> >
>> >
>> > This message has been scanned for malware by SurfControl plc.
>> > www.surfcontrol.com
>> _______________________________________________
>> Users mailing list
>> Users at openlayers.org
>> http://openlayers.org/mailman/listinfo/users
>
>
More information about the Users
mailing list