Hi,

Please try the following test

cd camino

track -inputmodel multitensor -inputfile test/tracking/crossing/crossing_05_0505_90.snr20.Bdouble.gz -seedfile test/tracking/crossing/rois/crossing_multi_roi.hdr > results.Bfloat

This uses some test data to do tracking with multitensor. It doesn't crash for me. Are you sure your input file is the right size? For two components, it should be

17 * 8 * N

bytes, where there are N voxels in the image. I normally name multitensor output according to the number of components.

Re: FA, there isn't a simple way to get the image in the multitensor case. I normally just run a standard tensor reconstruction and use that FA, because the multitensor FA is often quite unstable. I don't know if there's a generally accepted way to define an average voxel FA from a linear combination of tensors.