[vtkusers] Efficient Progressive Data Rendering?
Benny Bian
zbian at navitrak.com
Fri Apr 26 12:55:56 EDT 2002
Thanks John,
Few more questions. How do you decide which Actor is out of view port?
Do you have to modify all other Actors' world coordinates if you shift
the scene? Do I have to do all the display-view-world coordinates
translation to reassembly the scene?
Thanks lot.
Benny
-----Original Message-----
From: John Biddiscombe [mailto:jbiddiscombe at skippingmouse.co.uk]
Sent: Thursday, April 25, 2002 10:47 PM
To: Benny Bian; vtkusers at public.kitware.com
Subject: Re: [vtkusers] Efficient Progressive Data Rendering?
Benny,
<<<
I am working on a DEM visualization project using VTK. I successfully
read a small data set and build a poly data set to view. For a large
area of elevation data, I need progressively to create a small data set
(enough to fill the view window) and continuously load new data when
users continuously pan or fly to a different area. I searched the
previous posting and did not find the right approach. Can any one help
me?
I think constantly rebuild datasets from the beginning to view is very
expensive. How does VTK framework support this kind of job?
>>>
The way I've approached this sort of problem is to create a datareader
that
reads a tile of arbitrary size (n sq.km say). Then stuff lots of small
tiles
(say 4x4 grid for starters) into an assembly and create a separate actor
for
each sub tile. As the viewer moves over terrain, overload the assembly
update mechanism and discard any tiles that have moved out of view and
create new ones to fill in the spaces. This technique also has the
advantage
that textures can be overlaid on each tile (aerial photos/coverage
maps/or
just map maps) and get nice results. As textures are limited on most
graphics cards to 2048x2048 or thereabouts, it can be hard to texture a
large region nicely using just one actor.
JB
[PS. If you ever wish to contract out development of this kind of
thing...I'm looking for new contracts.]
More information about the vtkusers
mailing list