Been learning to directly manipulate the "matrix3d" css3 property, in particular rotating and translating the matrix. And at first I thought I had my calculations wrong but then I used an online calculator for calculating rotation I found my calculations are correct.

When I run my code in Chrome the top matrix is what I get when I use the css "-webkit-transform: rotateX(2deg) rotateY(2deg);". The bottom matrix I generate with javascipt and its the one that I want to directly plug into "-webkit-transform: martrix3d()", but as one can see one number is in a different location (note m1[0][1] and m2[1][0]). Note: I haven't calculated the rotation onto the translate values yet, so ignore the bottom row of both matrixes
Code:
0.9993908270190958, 0.001217974870087876, -0.03487823687206265, 0,
0, 0.9993908270190958, 0.03489949670250097, 0,
0.03489949670250097, -0.03487823687206265, 0.9987820251299122, 0,
99.93908270190957, 100.06088018891836, 0.002125983043832047, 1,

0.9993908270190958, 0, -0.03489949670250097, 0, 
0.001217974870087876, 0.9993908270190958, 0.03487823687206265, 0, 
0.03487823687206265, -0.03489949670250097, 0.9987820251299122, 0, 
100, 100, 0, 1
So the question, is the matrix that the browser generated different from a standard graphic matrix? And if so where do I find the correct formulas for working with the CSS3 matrix.