Field3D v1.5

[Magnus Wrenninge]

We're approaching the v1.5 release and I will be merging the dev_v1.5 branch into master within the next few days.

Here's a summary of the new features in v1.5:

MIP Fields - These are analogous to MIP maps for images. Given an input field, a series of lower-resolution fields are generated which allows fast filtered lookups at render time. Both sparse and dense MIP fields are supported, and for sparse MIPs, the sparse read cache works the same as for regular sparse fields, so memory use can be kept to a minimum. An additional benefit to MIP fields is that it provides light-weight representations that can be used for visualization.

Field3D::makeMIP() - A function that converts DenseField and SparseField instances into MIP representations. The filtering is fully threaded and uses separable convolution for fast execution. Box, Triangle, Mitchell and Gaussian kernels are supported.

Field3D::match() - A pattern matching utility that simplifies the selection of fields to load by providing a "<name>:<attribute>" pattern with optional wildcards. For example, a user can request "*:density", which would load all density fields in a file, or "woody:*", which would load all the  fields with the name 'woody'.

I'll send out an email once the merge happens.

Magnus

[Drew Whitehouse]

Thanks for the update Magnus, looking forward to trying the new MIP maps.

-Drew

--
You received this message because you are subscribed to the Google Groups "Field3D dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to field3d-dev...@googlegroups.com.
To post to this group, send email to field...@googlegroups.com.
Visit this group at http://groups.google.com/group/field3d-dev.
For more options, visit https://groups.google.com/d/optout.

--
Drew Whitehouse
NCI VizLab

[haggi krey]

Hi,

just tried to compile field3d under windows and it seems that it uses a 
libraray called fnmatch which is not available under windows by default.
Is there a workaround for this problem?

[Magnus Wrenninge]

Hi Haggi,

I don't have access to a Windows machine, so I apologize for introducing this problem.

It looks like PathMatchSpec is the windows alternative. http://msdn.microsoft.com/en-us/library/bb773727.aspx

Could you try the following:

1 - replace the #include with Shlwapi.h

2 - Replace the two fnmatch lines:

             if (fnmatch(pattern.c_str(), s.c_str(), FNM_NOESCAPE) == 0) {

    becomes: if (PathMatchSpec(s.c_str(), pattern.c_str())) {

             if (fnmatch(pattern.c_str(), attribute.c_str(), FNM_NOESCAPE) == 0) {

    becomes: if (PathMatchSpec(attribute.c_str(), pattern.c_str())) {

Let me know if that does the trick.

Magnus

[haggi krey]

Indeed it now compiles if I link with the shlwapi.lib. Thanks for the fix. That was fast ;!

[haggi krey]

Another one:

the unittests do not link because the methods adjustedMIPFieldMapping and mipResolution are not found in the lib. I added the FIELD3D_API and now the unit tests link without errors.

[haggi krey]

I tried to test the f3dinfo with an old f3d file and it crashed with a hdf5 read error. The unittests and sparsefield_io worked fine and the I could read the test_file.f3d created by sparse_field_io.

So it seems that old f3d files can create a crash if we try to read it with the new version.

[Magnus Wrenninge]

Hi Haggi,

I'll update the API with the appropriate FIELD3D_API tags. I haven't had any issue reading older files, but let me some new checks to the unit tests. Old files should always be readable. I will get back to you once the test suite is updated.

Thanks,

Magnus

On Fri, Jul 11, 2014 at 3:54 AM, haggi krey <haggi...@gmail.com> wrote:

I tried to test the f3dinfo with an old f3d file and it crashed with a hdf5 read error. The unittests and sparsefield_io worked fine and the I could read the test_file.f3d created by sparse_field_io.

So it seems that old f3d files can create a crash if we try to read it with the new version.

--