Closed hastaluegoph closed 3 years ago
That part computes the color of a pixel by Barycentric coordinates. The Barycentric coordinates however do not change, that's why the code is the same for both. However the bfm_verts_color
for a particular component differs from the mean computation. Just print the values of bfm_verts_color
for the mean and then for each of the components inside the for loop. You will see that it changes. Alternatively print mean_point_color
and point_color
and you will see that they actually differ. Does that answer your question?
I think what might cause the confusion is that color
is an object that depends on the color_coeffs
. By setting values for color_coeffs
the value of color
changes. To get the value of color
then, one just writes color.r
. If you want to know more about how that works, color
is a Chumpy object.
yes, that answer my question! I just check the colors values, it does change every iteration by setting the color_coeffs. Thanks for your help and patience!
Thanks for a great work ! It's brilliant to transfer the uv basis of FLAME from BFM. I have a problem here, for calculate the point color, should it multiply by the color coefficient? I found that part is exactly same with the mean color calculation.