Hi Michael,
the vertices in the patch file won't necessarily be in order (in general they won't be), so we couldn't just get rid of the vno lines. I guess we could have come up with a format that had all the indices, then all the coords, but it's not the way I wrote it (sorry).
Not sure I understand your second question. The surfaces will contain about 140K of vertices. Some of those will not exist in the patch, depending on how many you cut out when you generate it, but the triangles that are retained in the patch should be complete - that is either all the vertices are in the patch or none of them are.
cheers Bruce
On Tue, 20 Sep 2011, Michael Waskom wrote:
Hi, Two issues related to patch files.
Issue number 1:
If I generate an ascii flat patch file like so:
$ bin_patch=fsaverage/surf/lh.cortex.patch.flat $ mris_convert -p $bin_patch lh.cortex.patch.flat.asc
And then inspect the file that gets created:
$ head lh.cortex.patch.flat.asc #!ascii version of patch fsaverage/surf/lh.orig. The 1st index is not a vertex number 149297 296943 1 vno=0 -60.013611 4.446770 0.000000 2 vno=1 -56.163204 88.759064 0.000000 3 vno=2 -112.137077 -18.977211 0.000000 4 vno=3 -34.436668 -72.297958 0.000000
This is sort of annoying if I want to read it into Matlab, etc. (actually Python) using routines that expect an array of numbers. Is there any way to suppress the intercoordinate output?
Issue number 2:
Even if I get around the text file formatting issues, I run into some problems when I try and generate meshes with the resulting data. It seems like the triangles component of the surface makes references to vertex ids that don't exist anymore. In other words, the triangles array contains values that are larger than the number of rows in the vertices array.
Obviously I'm doing things that aren't officially supported by Freesurfer there, but does that jive with your knowledge of how the patch files are written?
Thanks!
Michael