The last time this error came up, it was related to the inclusion of zero-vector, but non-zero b-value entries in the DW encoding. I find it difficult to properly get a feel for bvecs/bvals, their formatting makes it difficult to figure out which components relate to each other. I recommend you use mrinfo dwi.mif -dwgrad instead, that should provide a more digestible version of the same information, and would make it much easier to spot such a problematic line – any chance you could post that here?
I’d also encourage you to post the output of the following two commands:
mrinfo dwi.mif -shell_bvalue -shell_size
dirstat dwi.mif
Note that you’ll need to be running a recent version for the latter to work.
I have a suspicion that the b=0 volumes having an actual b-value of 77 s/mm² but zero direction vectors might be the problem… If that’s the case, you may have some luck adding the -config BZeroThreshold 80 or similar to your command invocation.