The final image size was simply calculated based on the numbers in the PAR article on Wikipedia. Before resizing there is anti-aliasing with NNEDI3, which doubles the image size vertically. I tried to do the final resize with and without keeping the side bars and since I couldn't see any difference in detail, I cropped them.
The basic color correction is inspired by Moth3r's method, but I didn't really calculate anything, except for the chroma shift, but those are both less then 1.0 so I only included them in the script as 'notes'. The rest of the tweaks were made by comparing the histogram and vectorscope of the disc's SMPTE pattern to those of the standard color bars generated by Avisynth. In the following images, the sharper readings are those of the standard bars, and the pale overlay represents my capture. Raw on the left, corrected on the right.
Both luma and chroma comparisons were made on the first frame of the color bars segment from Side C. I noticed that after the first frame luma contrast starts dropping, until a couple of seconds later it stops. Meanwhile the chroma stays the same. I did the tweaks with the first frame (highest contrast) because its white level corresponds with flash frames from the movie better. The next image shows the fall of contrast in the color bars pattern on the disc.
