Computes an approximate transformation of a row.
The transformation that we are approximating
maximum error measured in input pixels that is allowed in approximating the transformation (0.0 for exact calculations).
This algorithm does not garuntee all values will be within the error threshold of the exactly transformed values. In practice, as long as the line is roughly linear, it should be at least very close if over the error threshold away from the actual values.
Transform each point exactly.