gdalwarp fails with many global projections

[jluis]

Hi, Here are a couple of examples on how gdalwarp fails when projecting a global image

gdalwarp -t_srs +proj=sinu w.tiff w_sino.tiff Creating output file that is 1147P x 572L.

gdalwarp -t_srs +proj=eck4 w.tiff w_eck4.tiff Creating output file that is 1145P x 572L.

gdalwarp -t_srs +proj=vandg w.tiff w_vandg.tiff Creating output file that is 572P x 572L.

Joaquim Luis

[Markus Neteler]

broken sinusoidal projection result (converted to JPG due to trac file size limit)

[Markus Neteler]

Frank,

e.g., with +proj=sinu the equatorial zone is flipped horizontally. Reproduced with recent trunk.

Markus

[Even Rouault]

• r18035 /trunk/gdal/ (alg/gdaltransformer.cpp apps/gdalwarp.cpp ogr/ogrct.cpp): gdalwarp: better heuristics to guess output extent when projection from lat/long world extent to other world global projections (sinu, eck4, vandg, EPSG:3785,...); add a CHECK_WITH_INVERT_PROJ configuration option that can be set to TRUE to force OGRProj4CT::TransformEx to check that the result is correct by checking with the inverse projection. gdalwarp can auto-sets this variable when it detects that some points on the edge of the computed raster cannot be projected
• r18036 /trunk/autotest/utilities/ (data/w_jpeg.tiff test_gdalwarp.py): Add test for gdalwarp to global projections (#2305)