When specifying the input transformation via parameters, affineSymTensorVolume and affineScalarVolume somehow interpret them differently, as the example below shows:

**** affineScalarVolume ****
/usr/local/dtitk/bin/affineScalarVolume -in mean_rigid3_fa.nii.gz -out test.nii -theta 10 -z 13.5 -y -15
inFile = mean_rigid3_fa.nii.gz
outFile = test.nii
reading the buffer ... time consumed = 0.044041
converting the buffer ... time consumed = 0.056023
Transformation Parameters (angle in degrees) :
x = 0
y = -15
z = 13.5
theta = 10
phi = 0
psi = 0
xx = 1
yy = 1
zz = 1
xy = 0
xz = 0
yz = 0
trans (inverse) applied = [ 0.000, -15.000, 13.500]
[ 1.000, 0.000, 0.000; 0.000, 0.985, 0.174; 0.000, -0.174, 0.985]

**** affineSymTensor3DVolume ****
/usr/local/dtitk/bin/affineSymTensor3DVolume -in mean_rigid3.nii.gz -out test.nii -theta 10 -z 13.5 -y -15
inFile = mean_rigid3.nii.gz
outFile = test.nii
interpolation option is LEI
reorientOption = PPD
reading the buffer ... time consumed = 0.246249
converting the buffer ... time consumed = 0.251324
Transformation Parameters (angle in degrees) :
x = 0
y = -15
z = 13.5
theta = 10
phi = 0
psi = 0
xx = 1
yy = 1
zz = 1
xy = 0
xz = 0
yz = 0
trans (inverse) applied = [ 0.000, -25.619, 33.873]
[ 1.000, 0.000, 0.000; 0.000, 0.985, 0.174; 0.000, -0.174, 0.985]
backward resampling ...time consumed = 1.809
converting the buffer ... time consumed = 0.254
writing the buffer ... time consumed = 0.150

fixed. the setCenter call was commented out in the scalar implementation.