xjjk has put up his scripts at http://caligula.rhombic.net/~xjjk/OpenStreetMap/scripts/
gdal2tiles-mp.py is the multi-threaded version Rename-TMS-layout-to-Google-Layout.py creates a script that renames tiles. For your 6 hour run, what resampling did you use, how big was the source GeoTIFF (with or without overviews) and what was the highest zoom range done? > From: Tyler Gunn [mailto:[email protected]] > Subject: Re: [Talk-ca] GeoTiff in JOSM > > > I believe you can also have JOSM get files directly from your hard > > drive, but I'm not sure the syntax to do so on the Mac. > > I did a render with gdal2tiles and was able to get it working fine in > Merkaator. Something about the tile number origin being opposite in > JOSM. > > > Maptiler is essentially a graphical front-end to gdal2tiles. Xjjk's > > version is the command line version modified. My iMac broke awhile > > back so I'm not sure how easy/hard GDAL is to set up with python > bindings on OS X. > > Ah, okay. I'll have to drop Xjik a line and see if I can get a copy > of the modified command line file. It seems there WAS a version of > gdal2tiles tat was optimized for multi-core, but the author seems to be > charging for it. > > There is a pre-made gdal install package for OSX, so it was extremely > easy to get up. Click and install. > > > You'll still be limited by your CPU :P > > On a single core my 1.9GB image took around 6 hours. So not too bad but > faster would be nice. > > > What I did for testing was work on a small section downloaded > > separately and benchmarked with different image scaling methods. The > > three worth considering are nearest, antialias, and lanczos. Nearest > > is fastest, preserves sharp edges but has the worst quality. Antialias > > is decently fast, decent quality. Lanczos is the best quality, > > preserves edges but is extremely slow. > > I'll have to try modifying the render settings and see how it goes. > I'll probably want to get the parallelized version first though. > > THanks! > Tyler _______________________________________________ Talk-ca mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk-ca
