Densitometry: how to determine the base of curves (e.g. in ImageJ) - (Feb/12/2016 )

I’m trying to quantify the differences in efficiency of cutting by different sgRNAs targeting a region of a gene using ImageJ to quantify bands following T7E1 digestion of heteroduplexes. According to Chapter 15 of a book called METHODS IN MOLECULAR BIOLOGY and a book I received from Sigma Aldrich, the following is an appropriate way to isolate the areas under the curves but I disagree and would like to understand the reasoning for the recommendation by Sigma and others.

Figure 1

Sigma et al. says to draw a line under all of the curves to determine the total amount of DNA loaded in the lane. The quantity of cleaved DNA is then determined from the smaller peaks (fig 1). 

Figure 2

Figure 3

The approach in fig 1 seems “unfair”. I think either both total DNA and the digested DNA should be measured to the bottom of the background (fig 2) or neither should (fig 3).

The smearing in Sigma's example seems to come from the endonuclease activity of CelI/Surveyor nuclease and isn't really background. It's not even an issue with T7E1 so my query doesn't really matter.