I can’t figure what you mean by this. If I have a video that is (vertcal) 1080x1920 and fit it into a video mode of 1920x1080 then the vertical length (1920) has to fit in 1080 pixels and the horizontal length has to shrink accordingly. Which basically means that the video is downscaled by the proportion 1920:1080. Downscaling discards information (pixels).
The comparison shots clearly show that increasing the resolution first (I calculated wrongly the 2880 should be 3414) creates less information loss than downscaling (reszie-to-fit-the-video-mode) does as shown by the following diagram:
Sorry, I do understand now - it has to do with using the “crop: source” filter, which I didn’t use
I must admit - the “Crop: Source” filter UI is one of the least intuitive UIs I have come across. Trying to understand it is not for the faint-hearted.