How did they built this customization feature on this site and which courses must I watch to learn how?

I don't know what you mean by "that's not how it was built". Perhaps you misunderstood my explanation. I'll go into a bit more detail:

Each view is a composite of several images of parts, all made from the perspective of a specific angle. There are 8 possible views, one for each major and secondary compass direction. The images are kept in separate folders representing each view direction, bike style, and chosen part color. For example, the image for a purple boy's saddle viewed when the bike is facing "south east" direction has this URL: .../parts/view8/boy/purple/saddle.png and looks like this (it's big so you'll need to scroll down a bit for the rest of the text): Notice how there is a good deal of blank space on the image. While each image is mostly transparent, it takes up the full space of the final composite, with the component already located at the proper position. This simplifies the combining process. And there are quite a few images! The script establishes event handlers for clicks on the controls, and then using the choices made combines the sub-images for each of the component parts to make up the total view.

All this keeps the script relatively simple but a significant effort was required from the graphics person(s) to create the large number of images that the script uses.

Does it make sense now?