[postgis-users] Old question resurfacing
Rémi Cura
remi.cura at gmail.com
Wed Dec 18 08:47:45 PST 2013
Hey,
getting something similar to a surface visualization out of points without
computing explicit surface is actively researched.
Usually it's very easy for a graphic card (GC) to drape a texture over a
geometry. It is its main purpose. I wouldn't worry too much about that
(already done in WebGL). Of course it would be cool to get some color out
of raster, maybe with postgis raster (I just began to test it, actually
trying to do the inverse : create DEM from points).
What you call average vector is the estimated normal vector of the point.
It can be estimated with various methods and is required for almost
everything (including display to get shadows or lighting effects very
useful to better "understand" the surface for an human eye).
If it's just for visualization, you can let the GC do the work. It is
incredibly fast. (various "shaders" do it).
If you need normal for other computing (most algorithm use it), you may
want to pre-compute.
This can become tricky for 3 reasons :
_it is in fact a local surface interpolation. So, what surface do you
choose, and how local is it (plan, sphere, polynom , fixed distance, fixed
number of neighbors ...)
_the computing always involve nearest neighbors. Needless to say you
**have** to use a LOD (for example, compute per patch, on small patch, or
use an octree).
_How to store it? You can do it in pointcloud of course but is is a little
boring. A dirty workaround would be to store it as float[3][]. Usually in
point cloud software it is stored like X,Y,Z and other attributes,
something like n_x,n_y ...
If you want to play with point cloud and see various option to visualize
with normal based lighting, and to compute normal with various parameters,
a very good open source software is CloudCompare<http://www.danielgm.net/cc/>.
It is easy (like drag and drop point cloud file) and powerful.
Cheers,
Rémi-C
2013/12/18 Basques, Bob (CI-StPaul) <bob.basques at ci.stpaul.mn.us>
> Rémi,
>
>
>
> What are your thoughts about colorizing points, from say aerial photo’s?
> Do you think this is possible as an on the fly piece of the equation, or
> something that would need to be stored in the database? I ask, because a
> lot of 3D rendering clients have difficulties with displaying stacked
> images as drapes on a surface, if there were a method to colorize the
> points on the fly, then multiple colorizing passes would have the same
> effect as stacking the source images onto a surface, which with densely
> packed points, might negate the need to even build a surface model at all.
>
>
>
> Also, have you thought at all about how to store average vectors for each
> point (a vector perpendicular to a plane created from nearest point
> averaging) Probably too soon to bring some of this up, but I thought it
> might be better up front than later.
>
>
>
> Bobb
>
>
>
>
>
>
>
> *From:* postgis-users-bounces at lists.osgeo.org [mailto:
> postgis-users-bounces at lists.osgeo.org] *On Behalf Of *Rémi Cura
> *Sent:* Wednesday, December 18, 2013 2:01 AM
>
> *To:* PostGIS Users Discussion
> *Subject:* Re: [postgis-users] Old question resurfacing
>
>
>
> Hey,
>
> I contacted the developper of XB PointStream specifically to ask him how
> to plug it to a postgres DB.
> It is no more an active project and he advised against it.
>
>
>
> At the other hand, with a much better licence (FreeBSD), active dev, and
> build in level Of Detail (LOD) feature (scene graph):
>
> http://potree.org/
>
>
>
> I asked the same question and it would be possible to plug it to db.
>
> We only need to develop a LOD inside DB to to make it possible to use it
> with huge database (I'm talking about 10's of billions at least in base,
> and several millions in cache/browser). ( Here my gothub project doc, the
> project is not clean enough to be used by others, but the slides contains
> some images :
> https://github.com/Remi-C/LOD_ordering_for_patches_of_points/tree/master/doc
> )
>
>
>
> Current WebGL limit with most graphic card seems to be around 10^6 points,
> it is very small, hence the obligation to have LOD support to have maybe 10
> * 10^6 points in cache in LOD structure in browser, and display only 10^6
> points, but cleverly :
>
> more points close to viewer camera and in front of it, few points when far
> from viewer camera, zero point behind camera.
>
> These are very old and effective computer graphics tricks.
>
>
>
> Cheers,
>
> Rémi-C
>
>
>
>
>
>
>
>
>
> 2013/12/17 Basques, Bob (CI-StPaul) <bob.basques at ci.stpaul.mn.us>
>
> Hi Frank,
>
> I've been watching this approach develop over the past few year. Some
> intereing approaches to using the point based methods of display that I've
> zeroed in on use feature that they describe as Surfels and/or Splats. Just
> Google those two key words and you'll see what I'm talking about.
>
> http://graphics.ucsd.edu/~matthias/Papers/Surfels.pdf
>
> One drawback to these approaches is that they are sort of a closed system
> in that they concentrate on only displaying the point based features. In
> the end, as a real world tool, I'll need to incorporate into this type of
> visualizers, I means of mixing feature types. Surface tins, CAD models,
> and Points. Having the points reside in a DB will both help with the
> integration as well as allow some processing of things to be accomplished
> on the fly in this mixed feature viewer.
>
> Bobb
>
>
>
>
>
> -----Original Message-----
> From: postgis-users-bounces at lists.osgeo.org [mailto:
> postgis-users-bounces at lists.osgeo.org] On Behalf Of Frank Henze
> Sent: Sunday, December 15, 2013 2:10 PM
> To: PostGIS Users Discussion
> Subject: Re: [postgis-users] Old question resurfacing
>
> I dont know, at which point the downsampling should take place.
> An interesting solution for webbased visualization could be XB Point
> Stream:
> http://zenit.senecac.on.ca/wiki/index.php/XB_PointStream
>
> Frank
>
> Am 15.12.2013 00:45, schrieb Rémi Cura:
> > I guess everybody has reached the same conclusion :
> > using pointcloud is is possible to manage 100's of billions of points
> > and retrieve quickly those of interest.
> >
> > Now most of the usage requiere severe downsampling, which isn't a part
> > of anything yet but may come (I'm working on adding this to pointcloud).
> > .
> >
> > Cheers,
> >
> > Rémi-C
> >
> >
> > 2013/12/12 Basques, Bob (CI-StPaul) <bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>>
> >
> > Frank,
> >
> > I don't know about a group, or for that matter centering it on just
> > PostGIS, but I see 3D as the next big Mapping product line that
> > people are going to be looking for. PostGIS is a piece in a bigger
> > puzzle. I've been researching different methods for storing and
> > segmenting the point cloud data on the server side for a while now
> > and PostGIS has percolated up to near the top of the list.
> >
> > There is a big piece related in how to generalize data for 3D
> > scaling in the browser, that's the biggest shortcoming I've become
> > aware of in order to make something run nicely in the browser, I've
> > also of late been forcing myself to think mobile, which has it's own
> > set of criteria as far as the browser goes.
> >
> > I would likely join the list of something call PostGIS 3d, but I
> > don't know that there is enough interest for a whole community
> > (yet), maybe it's better to stay ahead of the curve though.
> >
> >
> > Bobb
> >
> >
> >
> > -----Original Message-----
> > From: postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > [mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>] On Behalf Of Frank
> Henze
> > Sent: Thursday, December 12, 2013 1:17 PM
> > To: PostGIS Users Discussion
> > Subject: Re: [postgis-users] Old question resurfacing
> >
> > Hi all,
> >
> > briefly a few details to
> http://www2.htw-dresden.de/~s68071/3DWebGIS/:
> >
> > - Projektauswahl: Baalbek and Palatin are static X3D models (no
> > PostGIS)
> > - Projektauswahl: W3DS is dynamically generated from a PostGIS 2.0
> DB
> > via a "GetScene"-request by Geoserver
> >
> > I think Baalbek is too large (too many triangles) and therefore
> > possibly leads to crash. Palatin and W3DS should work.
> >
> > The 3D window is a standard WebGL window based on X3DOM. So it
> > should run right stable.
> >
> > Once again my question:
> > Is there an interest in a PostGIS 3D group?
> >
> > Frank
> >
> > Am 12.12.2013 16:55, schrieb Basques, Bob (CI-StPaul):
> > > Frank,
> > >
> > > Very nice. It froze up for me pretty quickly, I tried both FF
> > and Chrome. Still looked good while it ran.
> > >
> > > This stability aspect is a big piece of being able to make
> > something that folks will want to use. Data segregation will be a
> > big piece moving forward I think.
> > >
> > > Bobb
> > >
> > >
> > >
> > > -----Original Message-----
> > > From: postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > > [mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>] On Behalf Of Frank
> > > Henze
> > > Sent: Thursday, December 12, 2013 6:04 AM
> > > To: PostGIS Users Discussion
> > > Subject: Re: [postgis-users] Old question resurfacing
> > >
> > > Hi Bobb,
> > > hi all,
> > >
> > > https://hub.sharedgeo.org/apps/x3d/ looks great!
> > >
> > > We have similar requirements for a 3d WebGIS.
> > > A first prototype you can find at:
> > >
> > > http://www2.htw-dresden.de/~s68071/3DWebGIS/
> > >
> > > For "Projektauswahl:" select "W3DS"
> > >
> > > and then select "Historische Gebäude" (Historical Buildings)
> > >
> > > If there is nothing to see, then press on the left side "Alles
> > anzeigen"
> > >
> > > We use the community buildt Geoserver incl. Web 3D service +
> > X3DOM + JS.
> > >
> > > Some of our problems:
> > >
> > > How to import 3D geometries into PostGIS?
> > > Which formats and interfaces (CAD, X3D)?
> > > Point clouds in PostGIS (also import of).
> > >
> > > Is there a 3D PostGIS interest-group?
> > > If not, should one established?
> > >
> > > Frank
> > >
> > > Am 11.12.2013 17:44, schrieb Basques, Bob (CI-StPaul):
> > >> All,
> > >>
> > >> Nothing that far along. Did a couple of proofs of concept so
> far,
> > >> I've done a couple of presentations on the Visualizer approach.
> We
> > >> tried a couple of different things, x3Dom, allover'js
> > >>
> > >> You can see some of them here ( some of the pages take a while
> to
> > >> load the data in the background, be patient):
> > >>
> > >> https://hub.sharedgeo.org/apps/x3d/ (these will generally need
> a
> > >> webGL enabled browser)
> > >>
> > >> These are purely intended as a test of just how much data could
> > >> easily be squished into the browser before if blows, so you
> > might experience
> > >> some failures. Ideally the data coming into these would be
> > segmented
> > >> via a SQL call to PostGIS Pointcloud sources.
> > >>
> > >> The last two in the list are using some point clouds cut from our
> > >> recent data collect at 8pt per sq meter for the City (6 billion
> > >> points in all), these are using about 300k points each for
> example.
> > >>
> > >> Bobb
> > >>
> > >> *From:*postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> [mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>] *On Behalf Of *Rémi
> > >> Cura
> > >> *Sent:* Wednesday, December 11, 2013 10:08 AM
> > >> *To:* PostGIS Users Discussion
> > >> *Subject:* Re: [postgis-users] Old question resurfacing
> > >>
> > >> I would be very interested to know any attempt to visualize 3D
> point
> > >> cloud from data base !
> > >>
> > >> We did the same but our solution is far from perfect.
> > >>
> > >> Bob, is you rporject public/open source, have you any paper/doc
> > about it ?
> > >>
> > >> Cheers,
> > >>
> > >> Rémi-C
> > >>
> > >> 2013/12/11 Basques, Bob (CI-StPaul) <bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>
> > >> <mailto:bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>>>
> > >>
> > >> Hmm,
> > >>
> > >> I'm working with the Minneapolis International Airport (MSP) on a
> > >> project, any chance that data is open/accessible enough to play
> > with?
> > >> This could tie directly into a project I'm already working on.
> > >>
> > >> Thanks
> > >>
> > >> Bobb
> > >>
> > >> *From:*postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> <mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>>
> > >> [mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> <mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>>] *On Behalf Of
> *Gerry
> > >> Creager - NOAA Affiliate
> > >> *Sent:* Tuesday, December 10, 2013 2:14 PM
> > >>
> > >>
> > >> *To:* PostGIS Users Discussion
> > >> *Subject:* Re: [postgis-users] Old question resurfacing
> > >>
> > >> Bob, all:
> > >>
> > >> I agree. I'll have to spend some time with pointcloud but it DOES
> > >> look very promising.
> > >>
> > >> Another application? Lidar. Pointed at the sky, not at the
> > ground (we
> > >> use 'em to determine cloud layers [ceiling] and sky cover at
> > airports
> > >> for aviation data...).
> > >>
> > >> Thanks, all!
> > >>
> > >> gerry
> > >>
> > >> On Tue, Dec 10, 2013 at 1:49 PM, Basques, Bob (CI-StPaul)
> > >> <bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>
> > <mailto:bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>>> wrote:
> > >>
> > >> Gerry,
> > >>
> > >> Remi's idea about using a point cloud may be spot on for your
> use.
> > >> It allows you to set a point cloud down to a revolution if need
> be,
> > >> which seems like what you are looking for.. If the data becomes
> too
> > >> massive for insertion into DB at real-time speeds, then you could
> > >> also separate this revolution into separate DB's as well, you
> could
> > >> separate a whole number of ways, by elevation, or quadrant, or .
> . .
> > >>
> > >> I'm very interested in visualization possibilities with something
> > >> like this being available in a database. We're doing some
> > similar db
> > >> 3d visualization stuff on some rather dense point clouds. Your
> data
> > >> once available could use the same visualizer.
> > >>
> > >> Bobb
> > >>
> > >> *From:*postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> <mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>>
> > >> [mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> <mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>>] *On Behalf Of
> *Gerry
> > >> Creager - NOAA Affiliate
> > >> *Sent:* Tuesday, December 10, 2013 1:41 PM
> > >> *To:* PostGIS Users Discussion
> > >> *Subject:* Re: [postgis-users] Old question resurfacing
> > >>
> > >> Bob
> > >>
> > >> At least preliminarily, I can post-process, so speed of db adds
> > isn't
> > >> too troubling. Maintaining accurate representation of the
> bin-volume
> > >> data is, however, important.
> > >>
> > >> Typical rotation is 1-3 RPM, and a complete volume scan takes
> > ~11 min
> > >> in clear air (where you best see biologicals if so inclined) or
> ~5
> > >> min in one of the storm data collection modes. These are for
> common
> > >> WSR88D, stationary radars. SMARTR's and others we have here that
> are
> > >> mobile present a whole host of other options/data eval and speed
> > problems.
> > >>
> > >> Current radar data are nominally considered to have a horizontal
> > >> resolution of ~250 m, ignoring distortion or keyholing due to
> > >> range.Typically 16 elevations are scanned, once or or twice in
> storm
> > >> mode and a few less elevations in clear air mode.
> > >>
> > >> Now, the interesting thing that's on the horizon is Phased Array
> > Radar.
> > >> When that happens, more data, more resolution, and faster
> updates.
> > >>
> > >> gerry
> > >>
> > >> On Tue, Dec 10, 2013 at 11:05 AM, Basques, Bob (CI-StPaul)
> > >> <bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>
> > <mailto:bob.basques at ci.stpaul.mn.us
> > <mailto:bob.basques at ci.stpaul.mn.us>>> wrote:
> > >>
> > >> Gerry,
> > >>
> > >> Seems like the biggest hangup would be in adding the data to the
> DB
> > >> fast enough. How many points, per revolution, and what is the
> > >> frequency of a revolution (stationary Radar, correct, although
> as I
> > >> think about it, it could be mobile if needed, just need to add
> > in the
> > >> radar location to each record)?
> > >>
> > >> Bobb
> > >>
> > >> *From:*postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> <mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>>
> > >> [mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>
> > >> <mailto:postgis-users-bounces at lists.osgeo.org
> > <mailto:postgis-users-bounces at lists.osgeo.org>>] *On Behalf Of
> *Gerry
> > >> Creager - NOAA Affiliate
> > >> *Sent:* Tuesday, December 10, 2013 10:52 AM
> > >> *To:* PostGIS Users Discussion
> > >> *Subject:* [postgis-users] Old question resurfacing
> > >>
> > >> I asked this years ago, and I think Paul was less than pleased
> with
> > >> me (:-), but:
> > >>
> > >> Has anyone, in the ensuing years looked at encoding radar data
> > into a
> > >> postGIS database? We've a little idea that might benefit one
> > project,
> > >> and getting the radar data into a good geospatial format would be
> > >> beneficial.The data, of coure, would start out as
> > radial-distance and
> > >> intensity from the radar site, although we could preprocess it
> > by gridding.
> > >>
> > >> Thanks, Gerry
> > >>
> > >> --
> > >>
> > >> Gerry Creager
> > >>
> > >> NSSL/CIMMS
> > >>
> > >> 405.325.6371 <tel:405.325.6371 <405.325.6371>>
> > >>
> > >> ++++++++++++++++++++++
> > >>
> > >> "Big whorls have little whorls,
> > >>
> > >> That feed on their velocity;
> > >>
> > >> And little whorls have lesser whorls,
> > >>
> > >> And so on to viscosity."
> > >>
> > >> Lewis Fry Richardson (1881-1953)
> > >>
> > >>
> > >> _______________________________________________
> > >> postgis-users mailing list
> > >> postgis-users at lists.osgeo.org
> > <mailto:postgis-users at lists.osgeo.org>
> > <mailto:postgis-users at lists.osgeo.org
> > <mailto:postgis-users at lists.osgeo.org>>
> > >> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> > >>
> > >>
> > >>
> > >> --
> > >>
> > >> Gerry Creager
> > >>
> > >> NSSL/CIMMS
> > >>
> > >> 405.325.6371 <tel:405.325.6371 <405.325.6371>>
> > >>
> > >> ++++++++++++++++++++++
> > >>
> > >> "Big whorls have little whorls,
> > >>
> > >> That feed on their velocity;
> > >>
> > >> And little whorls have lesser whorls,
> > >>
> > >> And so on to viscosity."
> > >>
> > >> Lewis Fry Richardson (1881-1953)
> > >>
> > >>
> > >> _______________________________________________
> > >> postgis-users mailing list
> > >> postgis-users at lists.osgeo.org
> > <mailto:postgis-users at lists.osgeo.org>
> > <mailto:postgis-users at lists.osgeo.org
> > <mailto:postgis-users at lists.osgeo.org>>
> > >> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> > >>
> > >>
> > >>
> > >> --
> > >>
> > >> Gerry Creager
> > >>
> > >> NSSL/CIMMS
> > >>
> > >> 405.325.6371
> > >>
> > >> ++++++++++++++++++++++
> > >>
> > >> "Big whorls have little whorls,
> > >>
> > >> That feed on their velocity;
> > >>
> > >> And little whorls have lesser whorls,
> > >>
> > >> And so on to viscosity."
> > >>
> > >> Lewis Fry Richardson (1881-1953)
> > >>
> > >>
> > >> _______________________________________________
> > >> postgis-users mailing list
> > >> postgis-users at lists.osgeo.org
> > <mailto:postgis-users at lists.osgeo.org>
> > <mailto:postgis-users at lists.osgeo.org
> > <mailto:postgis-users at lists.osgeo.org>>
> > >> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> > >>
> > >>
> > >>
> > >> _______________________________________________
> > >> postgis-users mailing list
> > >> postgis-users at lists.osgeo.org <mailto:
> postgis-users at lists.osgeo.org>
> > >> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> > >>
> > > _______________________________________________
> > > postgis-users mailing list
> > > postgis-users at lists.osgeo.org <mailto:
> postgis-users at lists.osgeo.org>
> > > http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> > >
> > >
> > > _______________________________________________
> > > postgis-users mailing list
> > > postgis-users at lists.osgeo.org <mailto:
> postgis-users at lists.osgeo.org>
> > > http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> > >
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at lists.osgeo.org <mailto:postgis-users at lists.osgeo.org>
> > http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> >
> >
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at lists.osgeo.org <mailto:postgis-users at lists.osgeo.org>
> > http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> >
> >
> >
> >
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at lists.osgeo.org
> > http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> >
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
>
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
>
>
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20131218/f9cf68d7/attachment.html>
More information about the postgis-users
mailing list