Yes, sorry about that. My reply to your previous email follows, without the inline images, and I'll reply again to address your most recent email.
No, I haven't managed to solve this. I wasn't using JAI-EXT but a static call to JAIExt.initJAIEXT() in the test class has not changed the results. (I'm also using 23-RC, in order to assist with the testing effort.) The starting image, in +proj=eqc +lon_0=140.7 +ellps=WGS84, looks like the following (Australia in the middle longitudes; ±180° about 3/4 across from the left): https://anero.id/geotools-202003/himawari-8-visible-eqc.tif When I reproject to a full world web mercator tile using GeoTools, and the test case outlined earlier, I get a gap between 140.7°E and 180°E (i.e. the right hand side): https://anero.id/geotools-202003/geotools-web-mercator.png Regards, Andrew On Tue, 7 Apr 2020 at 04:22, Jody Garnett <jody.garn...@gmail.com> wrote: > I love seeing your test data, although I expect that only went to me as > the mailing list blocks emails past a specific size. > > Next I have a couple obvious questions: > > - I have not looked yet, did you reported this as a bug to the issue > tracker, having a test case (and test data) very much helps reproduce the > issue. > - Have you enabled advanced reprojection handling > <https://docs.geotools.org/latest/userguide/faq.html#q-what-about-raster-advanced-projection-handling> > (which is specially for this kind of problem) > > I am not familiar with this specific area of the code; but I am interested > in learning the resolution. > > -- > Jody Garnett > > > On Thu, 2 Apr 2020 at 15:59, Andrew Miskelly <amiske...@weatherzone.com.au> > wrote: > >> Hi Jody, >> >> No, I haven't managed to solve this. >> >> I wasn't using JAI-EXT but a static call to JAIExt.initJAIEXT() in the >> test class has not changed the results. (I'm also using 23-RC, in order to >> assist with the testing effort.) >> >> The starting image, in +proj=eqc +lon_0=140.7 +ellps=WGS84, looks like >> the following (Australia in the middle longitudes; ±180° about 3/4 across >> from the left): >> >> When I reproject to a full world web mercator tile using GeoTools, and >> the test case outlined earlier, I get a gap between 140.7°E and 180°E (i.e. >> the right hand side): >> >> Regards, >> >> Andrew >> >> >> On Fri, 3 Apr 2020 at 05:51, Jody Garnett <jody.garn...@gmail.com> wrote: >> >>> I was curious if you resolved this issue? Some care has been taken with >>> JAI-EXT operations to better handle no data etc... can you confirm your >>> system is setup to use these operations? >>> JAI itself has some ideas for extending the border area to prevent >>> artifacts near edges ideally a solution would hook into this. >>> -- >>> Jody Garnett >>> >>> >>> On Mon, 16 Mar 2020 at 21:13, Andrew Miskelly < >>> amiske...@weatherzone.com.au> wrote: >>> >>>> Hi GeoTools people, >>>> >>>> I have a pipeline for data from geostationary satellites which involves >>>> the following: >>>> >>>> - use GDAL to reproject the data from its native, geostationary >>>> projection to a cylindrical projection >>>> - use GDAL to arrange the data into a cloud optimised GeoTIFF >>>> - use a GeoTools based server to render imagery from the COG to >>>> clients >>>> >>>> The reason for reprojecting is that I've never been able to get >>>> GeoTools to transform from the geostationary projections. >>>> >>>> A basic geographic projection is fine in cases where the satellite is >>>> centred nearer to 0°E but when its field of view encompasses the dateline >>>> it's helpful to use an equidistant cylindrical projection that's centred on >>>> the same longitude as the satellite. That avoids dealing with an image >>>> which spans from 180°W to 180°E and comprises mostly "nodata" pixels (with >>>> the data at the left and right edges). >>>> >>>> When rendering the latter images using GeoTools, though, I get a band >>>> of missing data near the dateline. I'm wondering whether anybody can see >>>> where I'm going wrong. >>>> >>>> To reproduce: >>>> >>>> The following is an image produced from Himawari-8's visible red band, >>>> in the native geostationary projection. >>>> https://anero.id/geotools-202003/himawari-8-visible.tif >>>> >>>> Here it is after reprojecting to +proj=eqc +lon_0=140.7 +ellps=WGS84 >>>> using GDAL. >>>> https://anero.id/geotools-202003/himawari-8-visible-eqc.tif >>>> >>>> And here's the image from GeoTools, rendered using the code below. Note >>>> the band of missing data between about 140°E (the central meridian of the >>>> projection?) and the 180°E (the right hand side of the image). >>>> https://anero.id/geotools-202003/geotools-web-mercator.png >>>> >>>> public void test(File in /* himawari-8-visible-eqc.tif */, File out >>>> /* geotools-web-mercator.png */) { >>>> AbstractGridFormat format = GridFormatFinder.findFormat(in); >>>> AbstractGridCoverage2DReader reader = format.getReader(in); >>>> >>>> Layer rasterLayer = new GridReaderLayer(reader, >>>> >>>> SLD.wrapSymbolizers(CommonFactoryFinder.getStyleFactory().getDefaultRasterSymbolizer())); >>>> >>>> MapContent mapContent = new MapContent(); >>>> mapContent.addLayer(rasterLayer); >>>> >>>> StreamingRenderer renderer = new StreamingRenderer(); >>>> //Map<Object, Object> rendererParams = new HashMap<>(); >>>> //rendererParams.put(StreamingRenderer.CONTINUOUS_MAP_WRAPPING, >>>> Boolean.TRUE); >>>> //renderer.setRendererHints(rendererParams); // had no effect >>>> renderer.setMapContent(mapContent); >>>> >>>> BufferedImage image = new BufferedImage(2048, 2048, >>>> BufferedImage.TYPE_INT_RGB); >>>> try { >>>> renderer.paint(image.createGraphics(), >>>> new Rectangle(image.getWidth(), image.getHeight()), >>>> new ReferencedEnvelope( >>>> -20026376.39, 20026376.39, -20026376.39, >>>> 20026376.39, CRS.decode("EPSG:3857"))); >>>> >>>> ImageIO.write(image, "png", out); >>>> } catch (Exception e) { >>>> e.printStackTrace(); >>>> } >>>> } >>>> >>>> Many thanks in advance for any help. >>>> >>>> Regards, >>>> >>>> Andrew >>>> >>>> _______________________________________________ >>>> GeoTools-GT2-Users mailing list >>>> GeoTools-GT2-Users@lists.sourceforge.net >>>> https://lists.sourceforge.net/lists/listinfo/geotools-gt2-users >>> >>>
_______________________________________________ GeoTools-GT2-Users mailing list GeoTools-GT2-Users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/geotools-gt2-users
