The way to calculate BBand

Hi, all.
From the notebook, I found that the way calculates the band is a little confused.

prices['bb_up'] = prices.bb_high.sub(np.log1p(prices.close))
prices['bb_down'] = np.log1p(prices.close).sub(prices.bb_low)

Why we don’t use bb_high directly?

One is calculating the BB value. The result of this is calculating the difference between the value of the BB and the log close price.

One stationary and one not. Big difference in variables.