It's unlikely that the filters will each focus at the same place, so it is better to image a batch of one colour, then refocus and take another batch of another colour and finally the third colour. With an autofocuser, or no sleep, you could capture the set of RGB data in one night. I have no way of autofocusing so I accepted a compromise and cylce through the filters on each sub. This meant sub-optimal focus point on all colours - I mean it's close, but not perfect. I use a luminance layer, captured at the same time on a different scope, so I can just about get away with just so slightly iffy RGB data as the detail is in the luminance. However it results in larger stars, often with coloured fringes, so it's not something I wish to continue doing. I think there comes a point when you start doing mono + filter imaging, that an autofocuser is a key ingredient to the mix.