Same here. There's so much great component based modeling, I don't see the reason to use a capturing unit myself - right now with the exception of a Kemper Toaster, but that's got nothing to do with the captures but because I just find to be ergonomically excellent.
Would possibly change for me in case I could really capture pretty much anything with very little effort. As in creating some wicked nested parallel, frequency splitted and what not scenarios in my DAW (or also partially outside of it) and then just - *kazzzong* - capture it perfectly without much tinkering regarding training time/methods and such. But it seems that capturing tech isn't there yet.