home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.graphics.avs
- Path: sparky!uunet!spool.mu.edu!sdd.hp.com!usc!cs.utexas.edu!convex!convex!gardner
- From: gardner@convex.com (Steve Gardner)
- Subject: Re: Handling of "missing" or "special values"
- Sender: usenet@news.eng.convex.com (news access account)
- Message-ID: <1992Dec30.164201.10705@news.eng.convex.com>
- Date: Wed, 30 Dec 1992 16:42:01 GMT
- References: <1992Dec30.150955.18683@netnews.whoi.edu>
- Nntp-Posting-Host: imagine.convex.com
- Organization: Engineering, CONVEX Computer Corp., Richardson, Tx., USA
- X-Disclaimer: This message was written by a user at CONVEX Computer
- Corp. The opinions expressed are those of the user and
- not necessarily those of CONVEX.
- Lines: 45
-
- In article <1992Dec30.150955.18683@netnews.whoi.edu> rsignell@crusty.er.usgs.gov (Richard P. Signell) writes:
- >Is there a way to tell isosurface to ignore missing or special
- >values in a 3D grid? Setting missing values to zero or some
- >other value does *not* have the desired affect.
- You might try putting your data into a UCD. You will have to
- partition your data space into 3-D cells ( hexahedrons,tetrahedrons,
- pyramids, prisms) such that none of the nonexistent data points
- is included in a cell. Note that it must be a partition---no
- overlap is allowed. The ucd iso module should then do the right thing.
-
- By the way, the problem you describe is rather common with the
- AVSfield data type and one of the reasons why the UCD type is needed.
- Any time that the domain of definition of the function (the points
- at which functional values occur) is topologically "complex" it
- cannot be represented well with an AVSfield. AVSfields assume that
- the domain of definition is 2N-fold connected (where N is the
- dimensionality of the space) and that linear interpolation can be done
- between neighboring points. To make this more concrete, AVS fields
- in 3D assume that there are 6 neighbor points to point(i,j,k)
- {point(i,j,k+1),point(i,j,k-1), . . . point(i+1,j,k),point(i-1,j,k)}
- It also assumes this same neighbor relation for functional values
- and assumes that linear interpolation can generate the values
- between two points (e.g. data(i,j,k) and data(i,j,k+1)).
- Your functions do not fit this assumption. As you pointed out some
- other packages take care of topological problems like yours by
- allowing a complex data domain to be "gridded over" by a nice 2N-fold
- grid where "illegal" values are used to indicate "holes".
- This can be a problem since it is conceivable that any value you
- may choose to be illegal may exist in some data sets and it can
- also be wasteful of storage since the grid must be the bounding box
- of the data set. In AVS the use of UCDs to model complex topologies
- is the preferred method.
-
- I hope this answers your question. Feel free to contact me via
- email at the address below if you have any further questions.
-
- ==============================================================================
- Steve Gardner Convex Computer Corporation
- Software Engineer 3000 Waterview Parkway
- Scientific Visualization Applications P.O. Box 833851
- email: gardner@convex.com Richardson, Texas 75083-3851
- Phone: 214-497-4539 Fax: 214-497-4500
- ==============================================================================
-
-
-