Do Herbal A and Herbal B correspond to differences in imagery?

[MarcoP]

What makes the matter more complex, is that the situation for each section is different, but there is a general trend of decline as a section progresses. In the Zodiac folios, disappearance of color (among which a ton of red) goes hand in hand with simplification of the figures. 

I too have the impression of a "decline" or maybe "simplification". As you say, this is particularly clear in the zodiac, where the figures in Pisces are so highly elaborated with their costumes and patterned tubs.

A possible line of investigation about Herbal A and B could then be trying to quantify the degree of complexity of the images, e.g. by estimating the number of strokes in each painting. But it seems clear that this requires a considerable effort (of either manual counting or image processing) and it is doubtful that it could lead anywhere.

The colour annotations could be another element that was dropped in the process: since they appear on f1v, it seems they were there from the very start (assuming that 1r is the true first page as the red "initials" suggest).