Lat/Lon (WGS-84) -> Spherical Mercator. Errors introduced by forward/reverse projections seem very plausible, although the generate_tiles.py only seems to do a forward. I wonder if a better implementation of the GoogleProjection.fromLLtoPixel() method would do the trick?
Regards, Warren -----Original Message----- From: Jon Burgess [mailto:[EMAIL PROTECTED] Sent: 27 October 2008 18:31 To: Warren Vick Cc: [email protected] Subject: Re: [Mapnik-users] Batch generate_tiles.py - exacerbated tileoverlap On Mon, 2008-10-27 at 10:47 -0500, Warren Vick wrote: > I've enhanced the Mapnik generate_tiles.py script to generate larger > map > views and then output batches (8x8) of map tiles. I found this > necessary > as generate tile sets from large imagery file (say, 100MB per GeoTIFF) > resulted in a huge file load for each an every tile generated. The new > Python script is much quicker but I've exacerbated a problem... > > Some of you may have noticed a one pixel duplication from the bottom > of > one tile and the top of the tile below it. It's usually noticeable on > text (especially when rendered with a halo). My batch version of the > Python script makes this worse and I now see "stutters" of several > pixels (maybe 8 since that's my output tile grid size) which is very > noticeable. Does anyone know the reason for the problem? I thought it > might be a rounding error but Artem's first thoughts are that it's > projection related. We had something similar on the OSM map and that was fixed by changing the projection. What projections are you using for the map and underlying map data? IIRC the problem showed up when we projected a point forwards and then backwards. The result was different from the initial position. Jon _______________________________________________ Mapnik-users mailing list [email protected] https://lists.berlios.de/mailman/listinfo/mapnik-users
