02-07-2020 10:04 AM
it better to do the pansharpening using the TIL images or it's better to import the the two tils (usint the import tool) and then do the pansharpening?
The images are big so each steps require hours, I need to optimize.
After the pansharpenin I will use the control points tool
Solved! Go to Solution.
02-12-2020 09:44 AM
Here's what I'd personally do:
This way you'll gain the benefit of Spatial Modelers capability to multi-thread and perform the entire workflow in one step. If you have a GPU card and OpenCl configured the NNDiffuse will also be very fast.
But I'm biased toward Spatial Modeler :-)
02-13-2020 10:12 AM
I created the model:
Without the raster output before the PanSharpen I had this error:
13/02/20 19:02:07 SessionMgr(12728): Data was not read from the source. This is likely caused by stale pyramid information stored in the image.
13/02/20 19:02:07 SessionMgr(12728): Spatial model execution failed.
Should I create the pyramids?
Using the model in the image I have this output:
Min and Max are right? I aspected 0-10000
Furthermore I have many nodata (in red) in the image:
What do you think about? thanks!
02-13-2020 11:47 AM
Why use a bounding box for testing as you develop? A Preview is effectively a bounding box that will only process the area you're looking at.
Also, Rapid Atmospheric needs a large extent to build the parameters for the correction - how small of a bounding box are you defining? Or, preferably, don't. Just use a Preview instead. And no Raster Outputs (which are obviously going to cause inefficient processing) other than the final one.
This is working fine for me:
02-13-2020 11:52 AM
Oh, ,and if you're having a problem with "real" data being stretched below 0, getting truncated to 0 when converted to Int and then set to the NoData value when you output a file, try something like this:
02-13-2020 11:54 AM
Real-time atmospheric correction and pan sharpening. Can't get much more efficient than that!