Skip to main content

I'm reading and writing LAS files. There are warnings in the log files about X,Y,Z and being converted from Real64 to int32. How do I tell (with confidence) if this is important? Are the numbers being changed in a way that meaningfully changes the result?

LAS writer: Converting component 'x' from type 'Real64' to 'Int32'.  If this conversion is undesired, consider changing the type explicitly
LAS writer: Converting component 'y' from type 'Real64' to 'Int32'.  If this conversion is undesired, consider changing the type explicitly
LAS writer: Converting component 'z' from type 'Real64' to 'Int32'.  If this conversion is undesired, consider changing the type explicitly

https://knowledge.safe.com/questions/47749/combine-tilling-and-compress-las-files-without-con.html has a recipe for counter-acting the conversion, but it's not clear to me if that is needed. That discussion links to another page that says only int32 supported in the output format. If that's the case then why go through the extra work? 

I guess the question behind the question is: why is FME using Real64 at all if it's not supported?

ArcMap considers the points identical.

ArcMap before and after overlay of a LAS point. View scale 1:0.13. Distance from center to exterior of symbol is 0.006 meters. No measurable difference in point position.

 

0684Q00000ArCDTQA3.png

hi @maphew,

I don't think you should worry about this conversion. The coordinates are stored as int32 in LAS, but there is also scaling in the header, so you'll get the numbers you need at the end despite the conversion. We just warn that this conversion is happening.

FME itself supports real64 and there are other formats that support this data type. We don't want to limit FME to int32 only just because LAS was designed this way.

I hope this helps.

Dmitri

 

 

 


hi @maphew,

I don't think you should worry about this conversion. The coordinates are stored as int32 in LAS, but there is also scaling in the header, so you'll get the numbers you need at the end despite the conversion. We just warn that this conversion is happening.

FME itself supports real64 and there are other formats that support this data type. We don't want to limit FME to int32 only just because LAS was designed this way.

I hope this helps.

Dmitri

 

 

 

Thanks Dimitri.

An info message class would be better than warning in this instance. I spent a few hours trying to figure out what I should do about the warning. (A KB document explaining would have helped too.)


Thank Dimitri for the answer. I agree with Maphew it would be better to have the message class change from warning to info.


Reply