Selon Peter Hopfgartner <peter.hopfgart...@r3-gis.com>: > Hello Evan, > > I've tried this on your suggestion. What happens is, that all labels are > now much smaller, even harder to read then before. Basically, it seems > that tiles from a higher zoom levels are taken. This would be > reasonably, if there would be no labels on the tiles. But in the case of > the basemap.at tiles, labels are drawn on the tiles.
Did you try tweaking OVERSAMPLE_RATIO ? See http://mapserver.org/input/raster.html. Note: the doc is out of sync. The default value is 2.0 (not 2.5). > When working with tiles, even more if the tiles are from a third party, > the available options are much reduced, compared to WMS request. When > the raster information is lost due to a lower quality interpolation, > it's gone forever in the processing pipeline. > I did not have the time to look at the code yet (and probably I do not > have enough C++ experience either), I'll retry during the weekend. > > Regards, > > Peter > > On 09/18/2014 10:30 PM, Even Rouault wrote: > > Le jeudi 18 septembre 2014 22:23:06, Peter Hopfgartner a écrit : > >> Salut Evan, > >> > >> thanks for you recipe. Unfortunately, what I am really interested in is to > >> use this from MapServer. Could you give me a pointer in the source where > >> the resampling takes place? > > In gcore/rasterio.cpp : GDALRasterBand::IRasterIO(). I must warn you : > quite > > tricky code > > > > But perhaps you don't need code change. In MapServer did you try to define > > PROCESSING "RESAMPLE=BILINEAR" ? > > > >> Thanks, > >> > >> Peter > >> > >> Am 18.09.2014 um 22:11 schrieb Even Rouault <even.roua...@spatialys.com>: > >>> Le jeudi 18 septembre 2014 17:40:11, Peter Hopfgartner a écrit : > >>>> Hi, everybody > >>>> > >>>> we are using the GDAL WMS driver for getting maps from a public WMTS > >>>> service, http://www.basemap.at. Anyway, the resulting image does not > >>>> look smooth. In particular, I would guess that the the image was scaled > >>>> with something like a nearest neighbor algorithm. > >>>> So my question is simply, if I can switch the interpolation method in > >>>> some place. > >>>> > >>>> The XML file, taken from > >>>> http://gisforge.wordpress.com/2012/11/06/basemap-at-wmts-layer-in-qgis/, > >>>> is: > >>>> > >>>> <GDAL_WMS> > >>>> > >>>> <Service name="TMS"> > >>>> > >>>> <ServerUrl>http://maps.wien.gv.at/basemap/geolandbasemap/normal/google38 > >>>> 57/ ${z}/${y}/${x}.jpeg</ServerUrl> </Service> > >>>> > >>>> <DataWindow> > >>>> > >>>> <UpperLeftX>-20037508.34</UpperLeftX> > >>>> <UpperLeftY>20037508.34</UpperLeftY> > >>>> <LowerRightX>20037508.34</LowerRightX> > >>>> <LowerRightY>-20037508.34</LowerRightY> > >>>> <TileLevel>18</TileLevel> > >>>> <TileCountX>1</TileCountX> > >>>> <TileCountY>1</TileCountY> > >>>> <YOrigin>top</YOrigin> > >>>> > >>>> </DataWindow> > >>>> <Projection>EPSG:3857</Projection> > >>>> <BlockSizeX>256</BlockSizeX> > >>>> <BlockSizeY>256</BlockSizeY> > >>>> <BandsCount>3</BandsCount> > >>>> <Cache /> > >>>> > >>>> </GDAL_WMS> > >>>> > >>>> The command is: > >>>> > >>>> gdal_translate -srcwin 35500000 23500000 100000 100000 -of JPEG -outsize > >>>> 250 250 basemap_at.xml basemap_at.jpg > >>> Peter, > >>> > >>> Resampling done by gdal_translate is nearest neighbour, so not always > >>> that great. > >>> > >>> What you can do is select an outsize that is compatible with the source > >>> overviews. In TMS, the ratio between sucessive overviews work with factor > >>> of 2, so you should select an outsize of 10000 / 2^N. The closest is > >>> 10000 / 256 = 390.625 -> 391 > >>> > >>> So > >>> > >>> gdal_translate -srcwin 35500000 23500000 100000 100000 -outsize > >>> 391 391 basemap_at.xml tmp.tif > >>> > >>> and then: > >>> > >>> gdalwarp tmp.tif tmp2.tif -r lanczos -ts 250 250 (you can try different > >>> resampling kernel) > >>> gdal_translate tmp2.tif basemap_at.jpg -of JPEG > >>> > >>> Best regards, > >>> > >>> Even > > > -- > Peter Hopfgartner > R3 GIS > web : www.r3-gis.com > > -- Spatialys - Geospatial professional services http://www.spatialys.com _______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/gdal-dev
