In NASA's Scientific Visualization Studio we work with many Earth science datasets in different geographic projections that must be accurately positioned on the globe. Defining and mapping these by hand with individual manifold shaders is time-consuming and error prone. We developed an Interactive Data Language routine that automatically extracts the coordinate and projection information from geotif images. For each image, this routine writes out a Renderman Shading Language #include file passing the correct parameters to a projection routine that positions the related texture tile. The projection routine is a C++ Renderman plugin that computes the projection calculations in double precision. String formatted parameters to the projection plugin also allow double precision. We then create a "map-multiple" #include file referencing all the individual #include files in a tile set. This "map-multiple" file is added to an SLbox in SLIM so that the compiled shader references the appropriate tiles. We employed this method with high-resolution data for Greenland. From the Greenland Ice Mapping Project at Ohio State we obtained 30-meter topography data along with an ocean and an ice sheet mask. Each of these datasets consisted of a 6 x 6 array of 124 megapixel tiles. We also received seven sets of 20-meter Radarsat data from the Canadian Space Agency, mosaicked at the University of Washington's Applied Physics Lab. Each set consisted of a 5 x 5 array of 421 megapixel tiles. Using these tools, we generated 294 #include files to map this data: 108 for the topography data, the ocean and the ice sheet masks, 175 for the Radarsat brightness files and 10 for the "map-mulitple" files. The final master #include file includes all of the "map-multiple" files. With this method, we accurately mapped over 87 gigapixels of Greenland data.
from NASA's Scientific Visualization Studio: Most Recent Items http://ift.tt/1gok4rY
via IFTTT
No comments:
Post a Comment