sandworm: nodata expected behavior

[romain.janil]

Hi, what is actual expected behavior when generating with rasters containing nodata cells in sandworm?

1 Usecase : georef tiles non ovelapping of elevation data with nodata affect also mask layer (one raster source no nodata)--> nodata cells of elevation are to be found in masks cells, is this normal? I wasn't expecting source elevation to impact masks (haven't tried with albedo but maybe this is also the case)

2 usecase: is it possible actually to use nodata in rasters to composite rasters? My tests shows "corrupted"    or badly interpreted data over nodata locations...

[silent]

Hi Romain,

NODATA handle algorithm should be the same as it was for old Landscape tool.

 

Quote

2 usecase: is it possible actually to use nodata in rasters to composite rasters? My tests shows "corrupted"    or badly interpreted data over nodata locations...

NODATA is already should be converted to opacity, so NODATA == opacity == 0 and after that blending is done. It would be more obvious to us we can see your test data and perform import on our side as well :)

 

Quote

I wasn't expecting source elevation to impact masks

That's sounds odd, we didn't expect this also. Can you show us some screenshots of what's going on and expected results? It would be also great to check your original imported sources (at least part of them where NODATA and regular raster are overlapping).

Thanks!

[romain.janil]

Hi Silent,

here are some "early" tests snapshots:

here you can see elevation only (on waterglobal), where sea level at 0m is NODATA (-99999). It comes from a rasterization of vector convoluted with original elevation data

Global mask is a background layer (at 20mpx) covering all the extent, the purpose is to do some "mask by albedo" over all the terrain behind some additional landuse masks.

Weird thing is that :

here you can sea the debug mask (without the water) and see that the NODATA tags belonging to ELEVATION have created some black pixels in the MASK also: in original data these cells are white and the original data doesn't have any NODATA (it's a plain white raster). As I said all the landuse mask are lower density than the elevation.

In my test also, when I add an underlying raster (with obviously lower pixel density )  for the sea bottom, which should be visible under the NODATA of ground elevation, all elevation is mixed up here beteween NODATA and sea floor data...I'll try to post some imgs and share some data.