Correct. If you self a hybrid for enough generations, while simultaneously weeding out the specimens that diverge from your planned outcome, then you tend to eliminate the heterozygous alleles. So, for example, if 3 of those heterozygous genes are Aa, Bb and Cc (Cap = dominant; lower case = recessive), repeated selfing tends toward AA or aa, BB or bb and CC or cc. Which of these, and which combination of those three homozygous genes, will be selected by your own selection criteria of their phenotype: e.g. "I like the big, thick-leaf one that cures to black."
If disease resistance is on your list of criteria, then you need a basis for saying which exemplars are resistant or not resistant. But, for a beauty contest, just as with selecting tomatoes for durable shipping characteristics, you just keep on selecting for your criteria, until selfing is stable from one generation to the next. Typically, this takes 5 to 7 generations, with selection.
In practical tobacco terms, the Piloto Cubano (Puerto Rico) that I grew last year produced tall plants with longer, narrower leaves, as well as somewhat shorter plants with wider leaves. That seed batch was probably not a stable variety. (i.e. the seed had been collected from a plant--or several generations of plants--in Puerto Rico that had been exposed to pollen from some other variety.) I've decided to select for the shorter plant, larger leaf, in my growing of it in the future. I expect that to require at least a few generations to complete.
Bob