search

r-lidar / lidR

Airborne LiDAR data manipulation and visualisation for forestry application
https://CRAN.R-project.org/package=lidR
GNU General Public License v3.0
596 stars 131 forks source link

[2.1.5] lastransform doesn't update scale #304

Closed spono closed 4 years ago

spono commented 4 years ago

Ciao JR, I have a bunch of files to convert from EPSG 3035 to 32632. When reading them I get a warning:

Warning messages:
1: Invalid header: X scale factors should be factor ten of 0.1 or 0.5 or 0.25 not 1e-07 
2: Invalid header: Y scale factors should be factor ten of 0.1 or 0.5 or 0.25 not 1e-07

and indeed las@header shows: Scale factor X Y Z: 1e-07 1e-07 0.001

The transformation seems to work fine (coords are ok) but then when I try to write the las I get the related error: Errore: Invalid header: X scale factors should be factor ten of 0.1 or 0.5 or 0.25 not 1e-07

Is it something that the user should fix manually or what? I don't know if might be similar to #270 . Thanks in advance!

Jean-Romain commented 4 years ago

You have a precision of 10e-7 m in your data. It means that your points are located with a precision of a tenth of a micrometer (the size of a virus). This is actually valid and it should be writable in a LAS file but I hard coded a limited number of valid scale factors up-to 1e-4 (100 micrometers). Use lasrescale to set a reasonable scale factor.

The message should mention that ten of 0.1 means 0.1, 0.01, 0.001, 0.0001 but not 0.0000001

Jean-Romain commented 4 years ago

Also I really advice you to use LAStools for this kind of processing.

spono commented 4 years ago

You have a precision of 10e-7 m in your data. It means that your points are located with a precision of a tenth of a micrometer (the size of a virus). This is actually valid and it should be writable in a LAS file but I hard coded a limited number of valid scale factors up-to 1e-4 (100 micrometers). Use lasrescale to set a reasonable scale factor.

The message should mention that ten of 0.1 means 0.1, 0.01, 0.001, 0.0001 but not 0.0000001

oh, true, at a first glimpse I didn't realise it was that small, I thought it was related to the CRS scale (degrees) and would have been changed accordingly. [I don't know if I should ask here or in stackexchange, but] how should be identified a "reasonable scale factor" if it is not reported in (meta)data? (BTW thanks)

Jean-Romain commented 4 years ago

Well, in degrees seven digits can make sense. It is roughly a centimeter precision. I should maybe allow for such accuracy. Honestly I'm not an expert at this stage. If we are limited to 4 digits we can't store coordinates in degree. In another hand if I do not put strong constraint I can't handle case like your. A millimeter accuracy is way enough. But I think I will allow accuracy up to 10-7 for the case of data in degrees.

Jean-Romain commented 4 years ago

This issue should be fixed in rlas 1.3.6 https://github.com/Jean-Romain/rlas/commit/00384988c3e3a7845d36b5e29d47c9f593b6d6a3