[Liblas-devel] Broken links and wrong dates

[Ben Discoe]

On Fri, Mar 4, 2011 at 7:47 PM, Howard Butler <hobu.inc at gmail.com> wrote:
> I would say it's not a particularly good format for driving GUI applications (it is sequentially organized
> with no inherent spatial organization)

I suspect that's no problem, for the case of processing large raw data
(streaming operations on large files are generally sequential) or in
serving tiled datasets (which can be broken into a large number of
small files or any format, including LAS).  After all, PNG and JPEG
lack "spatial organization", but that doesn't prevent them from being
immensely useful and popular for serving huge spatially indexed
rasters ala Google Maps, Google Earth, and all their open brethren.

The case isn't perfectly analogous, because each pixel in a raster can
be located from its extents, whereas a point cloud needs a little bit
more coordinate system definition than that - although not much, it
could be a simple convention.

> Michael Gerlek and I are currently in the middle of bootstrapping something tentatively called libPC,
> which is where you'd want to add your own point cloud-ish format support to.

Yes, that's what i was imagining, depending on where my LAS testing takes me.

> visualization or real-time. If you want something on that end of the spectrum, I'd look at PCL
> (Point Cloud Library -- we sure suck at naming things) http://www.ros.org/wiki/pcl

I explored PCL (more specifically, it's file format "PCD") a couple
months ago, to the point of pulling out the PCD read/write code from
PCL and doing some testing.  I found PCD to be in a very "alpha"
state; the format has both ascii and binary flavors but i could only
find a single example file (ascii) (the "c-turtle") and little
geo-awareness in the API (or perhaps none, if my memory holds.)

> I hope that when libPC grows up, it can optionally fit into PCL as a data provider.

I also hope that PCL grows up. :)

-Ben