[Liblas-devel] Las2Oci problems

[T.J.F. Commandeur]

Thank you for the fast replies.

Martin:

My code is mostly taken from SPfinalizer but filtered only to what I needed.
Sharing the code is not a problem for me, but further development is not
possible because of my studies. The balanced point counts is not used but
only splitting into a quadtree and merging the cells into a MAXIMUM points
per cell due to the block size setting of Oracle. This maximum is based on
your minimum in SPfinalizer.

Howard:

No problem for finding the bug, I read through the source code of Las2Oci
but couldn't find wherein the problem lies. As noted above it is not
possible for me to create such a patch or collaborate on the libPC project
due to my studies. Besides that these could be some notes to keep in mind
while developing libPC? Looking at the projects I am happy to see that
people are developing open-source las/point cloud software!

Michael:

Of course I am willing to provide the source code. Mostly is copied/based on
SPfinalizer. To whom and where could I send the source to?

Kind regards,
Tom Commandeur

-----Oorspronkelijk bericht-----
Van: Michael P. Gerlek [mailto:mpg at flaxen.com] 
Verzonden: vrijdag 15 april 2011 18:26
Aan: 'Howard Butler'; 'T.J.F. Commandeur'
CC: liblas-devel at lists.osgeo.org
Onderwerp: RE: [Liblas-devel] Las2Oci problems

Following up on Howard's comments on libPC: I'd really like to have a
routine in there that creates a quadtree of based on spatial regions.  If
you are willing to put the code under a BSD license, I'd be willing to
integrate into the library.

-mpg


> -----Original Message-----
> From: liblas-devel-bounces at lists.osgeo.org [mailto:liblas-devel-
> bounces at lists.osgeo.org] On Behalf Of Howard Butler
> Sent: Friday, April 15, 2011 6:39 AM
> To: T.J.F. Commandeur
> Cc: liblas-devel at lists.osgeo.org
> Subject: Re: [Liblas-devel] Las2Oci problems
> 
> 
> On Apr 15, 2011, at 8:01 AM, T.J.F. Commandeur wrote:
> 
> > Hi all,
> >
> > For scientific research I created another kind of Lasblock based on
spatial
> finalization by M. Isenburg (spfinalizer).
> > This program is able to process larger point clouds and has a low memory
> footprint based on a quadtree structure.
> > When the program is done it creates either separate las files for every
> block or a las and kdx file.
> > Now these are created I want to load them to an Oracle spatial SDO_PC
> with Las2Oci.
> > I am using libLAS from OS4geo on windows, during the usage of Las2Oci I
> found some strange things.
> 
> Excellent!  I'm glad you found the code useful in some way.
> 
> > -          When writing las to a database using a kdx file Las2Oci tends
to
> duplicate the last block written to the database (kdx created by
Lasblock).
> 
> * Thanks for finding this.  It must be a bug.
> 
> > -          Las2Oci loads the complete kdx file into memory which seems
quite
> insufficient for memory usage. Can't the kdx be read line by line (block
by
> block) instead of completely at once?
> 
> * Yes, it could be made to do this. That it doesn't is because it was
going to be
> approached later when it became a problem.
> 
> > -          When executing Las2Oci without a kdx file but with multiple
las files it
> creates a new point cloud for every file. There is no option to add new
blocks
> to an existing point cloud instead when using multiple las files.
> 
> We would need a patch to fix this.
> 
> >
> > Is there a solution to these problems? Do I need to change my workflow
> for using Las2Oci to indeed add las files to an existing point cloud in
the
> database?
> 
> I have been working on the OCI stuff under a different framework that
> Michael Gerlek and I have been developing called libPC.  It's intent is
more of
> a "GDAL for Point Clouds" with things arranged in a processing pipeline.
This
> has a number of advantages over the libLAS OCI efforts, which are quite
tied
> to LAS. The lasblock code is available as a "filter" within this pipeline,
and it
> would be conceivable for you to make a filter using your modified
spfinalizer
> code as well if this is something you are interested collaborating on.
> 
> http://hg.libpc.org/main is where the code lives.  It is quite rough in
many
> ways, and there are definitely no prebuilt windows binaries that can
> consume those kdx files yet, but the basics of OCI reading and writing are
> there.
> 
> Howard
> 
> _______________________________________________
> Liblas-devel mailing list
> Liblas-devel at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/liblas-devel