[Liblas-devel] LASPoint::IsValid

[Howard Butler]

On Oct 2, 2009, at 2:42 AM, Volker Wichmann wrote:

> Hi,
>
> as there's some discussion on "invalid points" on the list at  
> moment, I like to ask a question about the IsValid check which I  
> have for some weeks now.

Is this using las2las?  You can use the --skip_invalid which despite  
its terrible name, means "skip the invalid check" (I think :) )

>
> IsValid checks (amongst others) the ScanAngleRank to be in the range  
> -90° to 90°. Our problem is now, that we have LAS files from MLS  
> (Mobile Laser Scanning) which have scan angle ranks ranging from  
> -128 to 128. So the check fails with this data.
>
> Does anybody of you know if the ASPRS is discussing this issue for  
> the next LAS specification and will relax this constraint (-90/90)?

http://lidarnews.com/open-source-software-licensing contains a lot of  
my thoughts on this subject, but I'll boil it down to this:  The  
specification is just a suggestion until there is a reference  
implementation.  libLAS isn't the reference implementation, and the  
ASPRS committee doesn't seem so keen on proclaiming one, so each  
vendor can pretty much do what they want and call it a "LAS" file (and  
they do, too.  Boy do they ever).

With regards to scan angle, the committee tried to outright through it  
out for the 1.3 release before being beat back by complaints.   
Evidently it isn't that useful.

As far as libLAS is concerned, I was willing to tilt at the windmill  
of validation earlier, but without the committee's sanction, it's just  
libLAS' word against whomever wrote the file.  Being strict  
contradicts libLAS' desire to be widely used in the industry,  
especially if we can't read everyone's crappy LAS.  To that end, I've  
taken on maintaining a sample library <http://liblas.org/samples>,  
which contains just about every file that a user comes on the list  
complaining that libLAS can't read ;)  My hope is the number of  
permutations of bad files with asymptotically approach zero as time  
goes by.  Of course, the committee makes geometrically more  
possibilities with every ridiculous (thank goodness 1.3 got scaled  
back from its original form) specification release :)

Howard