These are two sample fields along the ecliptic.  What are shown here are the luminosity 
function and the star chart for each field, which is 1 degree squared.  
The USNO catalog is complete to about 18 mag.  

The first one, f1024, is centered about ecliptic coordinates lambda=4, beta=0; 
i.e., it is a relatively less crowded field, with the number of stars 
in each magnitude bin of about several hundreds.  

The total number of stars brighter than 15 mag is 404, and the ratio of faint
(> 15 mag) to bright (< 15 mag, TAOS targets) stars is 0.24.  

The following is f1404, lambda=80, beta=0, very close to the Galactic anti-center, 
So this is a relatively crowded field.  The star density is notably higher, with 1482 stars brighter than 15 mag.  
The ratio of faint to bright stars is 0.27.  

This is the comparison of the luminosity functions of the 2 fields.

The density of faint stars is relevant because they contribute to sky background. 
Take these 2 fields as examples, as the number of bright stars increases, the number of faint stars increases proportionally 
(roughly 4 times).   If each trailed star image occupies say 300 pixels, then each TAOS 4 million-pixel frame should contain 
no more than 10,000 stars before background confusion starts to hurt.  

In other word, if N(< 15 mag) ~ 2000 per degree, then the total number of faint stars, N(> 15 mag) would be 4 times of this, 
and hence  in a 3 square degree TAOS frame, N(> 15 mag) would be about 24,000, already exceeding the confusion threshold.  

All faint stars do not contribute equally of course, but the lesson here is that there seems to be no lack ot fields with enough 
target stars, and more targets than necessary probably do not help.   The occultation simulator that Andrew and Tim have 
developed will be used to address just to what extent this statement is correct, particularly from one field to another.   If this 
turns out to be true, then a more sensitive CCD may not be necessary and observing with spectral filters is justified.