The question was: I use O 5.10.2 on an Indigo Elan. As anticipated, rs_fit values (using the option CONV) modulate as a function of an overall B-value applied through the command rsr_setup; however, the rs_fit value are not affected when I change individual atoms's B-values. Should I expect rs_fit to be sensitive to individual B-values ?

The answer: no, the dens is calculated using the overall B-factor defined in RSR_setup for *all* atoms (this also explains why B and RSCC are highly correlated, typically with a corr coeff of ~-0.8)

By the way, i don't think CONV or DIFF is relevant for RS-fit (only used in the real-space refinement to use sum of differences or convolution product as the target function)