Okay, I see, thank you for the answers!
But there is two problems with this method.

As testing I choosed M1 to H4 timeframes, and calculations for 400 bars:

Problem 01

I attached a plotted result. As you can see there are a lot of useless data here, because my broker do not even store this amount of M1 candles. But I need them for calculating the old H4 candles. These are easily avaliable of couse if I go by BarPeriod = 60*4, TimeFrame = 1. But the above way I can't get the old H4 bars, because these are calculated by Zorro with the M1 candles - wich are not exists.

(I also want to avoid that I mix the previously downloaded historical M1 data and my broker's data.)

Problem 02

First loading of months of M1 data is incredibly slow and unneccessary if I calculating with 60 assets.

Possible solution

A solution can be that I run multiple Zorro instances - each one with different BarPeriod, and not changing TimeFrame at all. These scrips could be fast and simple, but for this I need to make a messaging bridge between them - for example some TXT files or interprocess communication.

Have you got a better idea?
Maybe I miss something? Maybe there is a way to download H4 candle data even if my BarPeriod is 1?