Why the Line (antialiased) function of openCV2 gives different results on CV_16UC1 and CV_8UC1 without overflow

Question

themap = cv.CreateMat(8,8,cv.CV_8UC1)
cv.SetZero(themap)
cv.Line(themap,(0,0),(7,7),(10),1,cv.CV_AA) #draw a line
print np.asarray(themap[:,:])

#######output
[[ 0  0  0  0  0  0  0  0]
 [ 0  0  0  0  0  0  0  0]
 [ 0  2  7  2  0  0  0  0]
 [ 0  0  3 10  3  0  0  0]
 [ 0  0  0  3 10  3  0  0]
 [ 0  0  0  0  2  7  2  0]
 [ 0  0  0  0  0  0  0  0]
 [ 0  0  0  0  0  0  0  0]]

However, when I change the image type to cv.CV_16UC1, the result is totally different:

themap = cv.CreateMat(8,8,cv.CV_16UC1)
cv.SetZero(themap)
cv.Line(themap,(0,0),(7,7),(10),1,cv.CV_AA) #draw a line
print np.asarray(themap[:,:])

#######output
[[10  0  0  0  0  0  0  0]
 [ 0 10  0  0  0  0  0  0]
 [ 0  0 10  0  0  0  0  0]
 [ 0  0  0 10  0  0  0  0]
 [ 0  0  0  0 10  0  0  0]
 [ 0  0  0  0  0 10  0  0]
 [ 0  0  0  0  0  0 10  0]
 [ 0  0  0  0  0  0  0 10]]

I'm totally confused by such results. Since the pixel value is only 10, there should not be overflow problems for cv.CV_8UC1 image type. Why the results of cv.Line can be so different?

The opencv in installed in /usr/local/Cellar/opencv@2/2.4.13.6_2/lib/python2.7/site-packages/cv.py with brew install opencv@2.