[gdal-dev] LAS support

[Mateusz Loskot]

On 29 November 2013 10:32, Rui L. Pires <rlpires at gmail.com> wrote:
> On 28 November 2013 15:09, Mateusz Loskot <mateusz at loskot.net> wrote:
>> On 28 November 2013 13:56, Rui L. Pires <rlpires at gmail.com> wrote:
>> > On 28 November 2013 11:03, Mateusz Loskot <mateusz at loskot.net> wrote:
>> >> On 28 November 2013 09:51, Rui L. Pires <rlpires at gmail.com> wrote:
>> >> >
>> >> > Would there be any particular reason for not supporting it?
>> >>
>> >> Point cloud is not a vector, not a raster either? :-)
>> >
>> > That might be debatable, but it won't be me reopening that discussion.
>> > Having said that, if you could assure me that all formats supported in
>> > http://www.gdal.org/ogr/ogr_formats.html are vector formats, I would
>> > have
>> > understood your point (of not offering a LAS driver).
>>
>> I intentionally put question mark to indicate it's open question for me as
>> well.
>> The nature of LiDAR data might be confusing and I'm interested to
>> learn more about it.
>> Dale Lutz said, "point cloud is a badly behaved raster".
>> I (used to) say, it's not a vector.
>> But I still haven't found clear answer "what is it" really :)
>
>
> It's interesting you mention LiDAR, but I would like to stick to point
> cloud, or point data set if you will.

I meant point cloud. I sometimes may freely interchange LiDAR data and
point cloud data, not saying it's correct.

> I was approaching this issue and particularly OGR as a solution for data
> conversion and reduction. I am quite fond of the functionality OGRLayer
> makes possible in a transparent way, such as reprojection and spatial
> querying.

Those are valid use cases of course and I don't argue here.
Just, since OGR is an abstraction layer and point clouds are enormous,
I wouldn't expect high performance.
That's why, IMHO, point clouds require dedicated toolkits (PDAL, PCL).

Perhaps, you can try las2ogr to see how OGR in general will work for your uses.
Besides, writing OGR driver with libLAS (or PDAL) should be feasible.

Best regards,
-- 
Mateusz  Loskot, http://mateusz.loskot.net