[Tilecache] Strange artifacts (lines) across Google Map tiles

Roger André randre at gmail.com
Sun Nov 16 12:13:21 EST 2008


Apologies for repeating this posting, but I realize I probably didn't do a
good job of explaining the problem before.  Basically it appears that
Tilecache is making requests for tiles with extents that a WMS service
cannot fill under certain conditions.  The result manifests itself as weird
horizontal stripes across various parts of my Google Map, with the location
of the stripe varying depending on the Zoom level of the map.  I have posted
screenshots of the problem
here<http://lagerratrobe.blogspot.com/2008/11/strange-tilecache-artifacts.html>.


As one of the most noticeable stripes occurs at Zoom Level 2, I was able to
find the actual TileCache WMS request in my Apache logs.  It looks like
this:
http://localhost/cgi-bin/mapserv?map=/var/www/mapfiles/test/test.map&layers=imagery&srs=EPSG%3A900913&version=1.1.1&bbox=-20037508.34%2C-20037508.34%2C-10018754.1704%2C-10018754.1704&service=WMS&width=256&styles=&format=image%2Fpng&request=GetMap&height=256

When that request is made, the tile at the bottom of the sample page I've
linked to above is returned.  You can see that the bottom few rows of pixels
contain only background color, making the stripe that runs across the
example screenshots.

I figure there are a couple possibilities for why this problem is ocurring.

1) TileCache is specifying an incorrect bbox size when it makes a WMS
request for the tile.
2) The projection definition for epsg 900913 is incorrect in some way, and
the reprojection from 4326 to 900913 is flawed.

Either way, it makes it difficult to display rasters with large extents.
One element that I can eliminate though is the renderer.  The flaw is
reproducible using various versions of Mapserver, as well as Mapnik.

I would love to actually have a chance to discuss this problem, and will
gladly provide any additional information upon request.

Roger
--
Just in case someone else wants to see this for themselves, here are the
steps needed to replicate the condition I'm seeing.

Step 1.  Create a large rectangular image of Size 14400 x 7200.

Step 2.  Assign georeferencing to the image:
gdal_translate -a_ullr -180 90 180 -90 -a_srs "EPSG:4326" big_black.tif
wgs84_big_black.tif

Step 3. Load this image into some sort of WMS rendering system and assign a
background color value of red to it.

Step 4. Make sure you have the spherical mercator projection defined in your
epsg file.
<900913> +proj=merc +a=6378137 +b=6378137 +lat_ts=0.0 +lon_0=0.0 +x_0=0.0
+y_0=0 +k=1.0 +units=m +nadgrids=@null +no_defs <>

Step 5. Make a WMS request for a tile that replicates one that TileCache
makes:
http://localhost/cgi-bin/mapserv?map=/var/www/mapfiles/test/test.map
&layers=imagery
&srs=EPSG%3A900913
&version=1.1.1
&bbox=-20037508.34%2C-20037508.34%2C-10018754.1704%2C-10018754.1704
&service=WMS
&width=256
&styles=
&format=image%2Fpng
&request=GetMap
&height=256

Step 6. Admire the red line at the bottom of the image which indicates no
data.
--
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/tilecache/attachments/20081116/88301c86/attachment.html


More information about the Tilecache mailing list