sandworm generation problem

[romain.janil]

Hi in 2.16.1 I'm facing a weird problem with sandworm.

I have a big export area with some low res imagery (~15m) and several "insets" with high res (below 0.4m).

Generation is fine as long as I generate with only one inset on top of the low res one, but, if I generate with many, all the high res insets get merged into one single lmap (which make sense) but the imagery is totally mixed up (like rows of the imagery being passed trough a paper shredder :( ) I can't really share some pics unfortunately , is this a documented bug, I never got that problem, maybe because usually I stack imageries but there are only one per pixel density, even if many.

Thanks 

[romain.janil]

Just to be more specific everything is in the same source projection of course, I'll try to capture a snapshot of the problem I should be able to share in PM

[bmyagkov]

5 minutes ago, romain.janil said:

Just to be more specific everything is in the same source projection of course, I'll try to capture a snapshot of the problem I should be able to share in PM

Hello!

You can switch to debug mode through the "Helpers" menu by selecting "Opacity World Normal", which will allow us to see what happens to the heights without displaying albedo data to keep your data private.

Thanks!

[romain.janil]

hmm, the problem is specifically on albedo, not heights.

I tried to reproduce the mashed up problem regenerating with 2 high density images on top of unique low one, and this time....result is still buggy but not the same as before: I got two landscapeLayerMaps under my landscapeTerrain 0.495mpx lmaps (where before I got only one), pointing to the same file !

so I have the same imagery mapped over 2 different zones, but the sources imageries are obviously different. Sources images are georeferenced and have the same resolution (or density)...

[romain.janil]

see pic (background low res imagery is tinted differently to make visual clearer) (mosaic effect is done in photoshop :), low res albedo is not THAT low res :))

 

Edited April 20 by romain.janil

[romain.janil]

where it should be this (preview of the imagery layers in sandworm)

[romain.janil]

for the record, I also tried to clear sw cache (which took a very long time) once, and retried but result generation seems to be always buggy, in case of several separated sources images with exact same densities (at least that's what I'm suspecting, might be related to something else tho, because I've already generated some terrain with tiled sources imgs and everything went well, but it was with previous 2.14/2.15 versions) 

Edited April 20 by romain.janil

[bmyagkov]

@romain.janil What projection was used for the generation? Currently, we are not aware of any issues related to this. It would be great if you could send us the exact results in a private message, if possible.

Thanks!

[romain.janil]

Indeed, I'm trying to isolate the problem based on the assumption this might be related specifically to this projection(because also I never encounter this problem before)  (and eventually make some tiny unitary samples); I keep you posted, I think in PM.

 

[romain.janil]

OK found somehow workaround changing projection (but there might be specifically a problem with the one I used, as told in PM), for the moment with a limited subset of data, I'll test with more to see if issue doesn't surface again. For the moment, consider the problem solved. Thanks!

[werner.poetzelberger]

Just for curiosity. Are you using different projections in the sources?

[romain.janil]

Hi, no everything was in the same projection (the one which gave me bad output), all source data; sandworm project, output.

Now I'm using another one, everything is in that one projection , and now output is correct...

[werner.poetzelberger]

Which projection do you mean?
The projection of the sources? Of the Sandworm file or the reprojection?
And which projection do you use?

Just asking as I am working with sandworm as well ;)

Edited April 24 by werner.poetzelberger

[romain.janil]

HI, everything, :)

I used a specific local proj (which is mainly used topograpphy, surveys etc) for a certain part of the world which gave me bad output result: sources data was in that proj, the sandworm project was in that one, and the ouptut was also in that one.

Now I reprojected all source data in a more "generic" world projection (3857), set the sandworm project to 3857, output to 3857 and now output is ok, fortunately, but at the same time highly expected, since 3857 is not an exotic projection.

[werner.poetzelberger]

Ah Interesting.

I use a World Mercator EPSG3395 as Output. 
I saw that I needed to match the Sandworm and the input projection but I thought that the output is better to use a standard one.

Ha, still an unexplored continent for me, the GIS world..

Thanks for your info!

Best.

WErner