Well, firstoff, there is quite a long discussion of this process at www.photo.net with myself and Bujor B among the contributors. There are examples of quite a few types of process presented including shots taken in-camera.

The first developer does indeed influence the results more than any other step. I found that Dektol 1:2 or 1:3 to be best. I even mixed a no-bromide version and experimented with bromide level. Sulfite level in the color developer is also a significant contributor to reduction of contrast. And, the curves do not match as you see. There is crossover if it is not done properly. You get cyanish highlights which means that the cyan layer in that region was not developed properly in the first developer. Bromide seems to control this to some extent.