A
batched instance of size
n is a 3-tuple
(m, b, t), where
m is a
mesh,
b is a buffer of
n 4x4 matrices allocated on the GPU, and
t is an array of
n transforms allocated on the CPU. For each
i where
0 <= i < n,
b[i] is the 4x4 model matrix produced from
t[i]. The contents of
b are typically recalculated and uploaded to the GPU once per rendering frame.