Need to get rid of min and middle eigenvalues.

Color by orientation in 3D is more complicated.

It's odd that it works for lines

Antoine DUCHAMPS wrote:

hi all,

I have problems visualizing tensors as ellipsoids. They appear as B/W in
the 3D scene (in color in 2D views). The same problem with tubes. There
is no problem with lines. I'm sending you a snapshot of the problem.

I'm using the binary files of slicer 3.3.2 for linux 64.

Tensors are obtained from the 01053_case of tutorial_material.

Interestingly, if I load tensor.nrrd from the same tutorial I don't have
any problem.

Any ideas?

Antoine.

I've seen this happen with lines. Usually when I try to change visibility or other FiberTracts model properties in Model hierarchy, as opposed to using DisplayLoadSave

Possibly unrelated: the output of DistanceTransformModel is goofy too: model appears all black initially. I can make it take some color by changing one of specular/etc sliders, but the surface is never quite right.

So I am pretty sure there are two bugs, one with lookup tables (reported here that affects all glyphs including lines) and another with transformation matrices (that may happen in DistanceTransformModel as well as with some badly formed ras to ijk matrices in glyphs).

Lauren O'Donnell to slicer-users
show details 9:43 AM (31 minutes ago) Reply

I think it has to do with the ijk to ras matrix used to place the
glyphs in the scene. If the normals are pointing inwards
something similar to this can happen though I haven't seen totally
black glyphs before. But this problem would explain the difference
between the datasets and the types of glyphs affected (lines don't
have a surface or normals).

So, I think this bug is probably not related to the previously reported one.

I would recommend making sure the matrices in the nrrd header for the
tutorial make sense. You can also save the scene and view the ijk to
ras matrix as text in the mrml file.

Lauren

Closing. Slicer3 is not being developed anymore.