Folks:

We were reluctant to believe it, but we think that in 3dWarp the -matvec_xxx flags are backwards, or the docs are backwards.

There are several helps that mention the issue, all are consistent. For example: 3dTagalign's help says:

-matvec mfile ... can be used as input to the '-matvec_out2in' option of 3dWarp

(and from the docs of the actual in2out and out2in flag I can't tell what they are supposed to do.)

To test, we created some unambiguous geometry in a test volume and a slightly rotated version of same. Then we used this script:

-----------------------------------
3dTagalign -master RotTest_Ref+orig -matvec RotTest.mat -rotscl -prefix RotTest_3ta -verb RotTest_APn10_IS20+orig

3dWarp -matvec_in2out RotTest.mat -cubic -prefix RotTest_Ruth RotTest_APn10_IS20+orig

3dWarp -matvec_out2in RotTest.mat -cubic -prefix RotTest_Doc RotTest_APn10_IS20+orig
-----------------------------------

Ie: we had 3dTagalign perform an alignment/rotation (which looked good) then had 3dWarp repeat the same rotation with the matvec file. The out2in version (that follows the docs) rotated in the wrong direction, but the in2out version rotated in the correct direction getting the same results as 3dTagalign.

Are we missing something here?

Graham